feat: 3D audio spatial chat — volume based on distance (fix #1544)

- Add SpatialChatAudio component with distance-based volume rolloff
- Each agent gets distinct tone (freq + waveform)
- Stereo panning based on relative position to camera
- HRTF PannerNode support for voice chat streams
- Integrate with addChatMessage for spatial notifications
- Tests + docs included

.gitea/workflows/pr-backlog-monitor.yml

@@ -0,0 +1,108 @@
+name: PR Backlog Monitor
+
+# Runs every Monday at 06:00 UTC — fires an issue if any repo in the org
+# accumulates more than PR_THRESHOLD open PRs.
+#
+# Background: timmy-config hit 9 open PRs (highest in org) before triage.
+# This workflow catches future buildups early.
+# Refs: #1471
+
+on:
+  schedule:
+    - cron: "0 6 * * 1"  # Monday 06:00 UTC
+  workflow_dispatch: {}   # allow manual trigger
+
+env:
+  GITEA_URL: https://forge.alexanderwhitestone.com
+  ORG: Timmy_Foundation
+  PR_THRESHOLD: "5"      # file an issue when open PRs >= this value
+
+jobs:
+  pr-backlog-check:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Setup Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.x"
+
+      - name: Check PR backlog across org repos
+        env:
+          GITEA_TOKEN: ${{ secrets.GITEA_TOKEN }}
+        run: |
+          python3 - <<'EOF'
+          import json, os, sys
+          from urllib.request import Request, urlopen
+          from urllib.error import HTTPError
+
+          BASE   = os.environ["GITEA_URL"]
+          ORG    = os.environ["ORG"]
+          TOKEN  = os.environ["GITEA_TOKEN"]
+          THRESH = int(os.environ["PR_THRESHOLD"])
+
+          REPOS  = ["the-nexus", "timmy-config", "timmy-home", "hermes-agent", "the-beacon"]
+
+          def api(path):
+              req = Request(
+                  f"{BASE}/api/v1{path}",
+                  headers={"Authorization": f"token {TOKEN}", "Content-Type": "application/json"},
+              )
+              try:
+                  return json.loads(urlopen(req, timeout=30).read())
+              except HTTPError as e:
+                  return {"_error": e.code}
+
+          backlog = {}
+          for repo in REPOS:
+              prs = api(f"/repos/{ORG}/{repo}/pulls?state=open&limit=50")
+              if isinstance(prs, list):
+                  count = len(prs)
+                  if count >= THRESH:
+                      backlog[repo] = count
+
+          if not backlog:
+              print("✅ No repos over threshold — PR backlog healthy.")
+              sys.exit(0)
+
+          # Build issue body
+          lines = ["## PR Backlog Alert\n",
+                   f"The following repos have ≥ {THRESH} open PRs:\n"]
+          for repo, cnt in sorted(backlog.items(), key=lambda x: -x[1]):
+              lines.append(f"- **{ORG}/{repo}**: {cnt} open PRs")
+          lines += [
+              "",
+              "### Recommended actions",
+              "1. Review and merge ready PRs",
+              "2. Close stale / superseded PRs",
+              "3. Run `python3 scripts/pr_triage.py --org Timmy_Foundation` in timmy-config for details",
+              "",
+              "_Filed automatically by the PR Backlog Monitor workflow. Refs #1471._",
+          ]
+          body = "\n".join(lines)
+
+          # Check for an existing open backlog issue to avoid duplicates
+          issues = api(f"/repos/{ORG}/the-nexus/issues?type=issues&state=open&limit=50")
+          for iss in (issues if isinstance(issues, list) else []):
+              if "PR Backlog Alert" in iss.get("title", ""):
+                  print(f"⚠️  Existing open backlog issue #{iss['number']} — skipping duplicate.")
+                  sys.exit(0)
+
+          import urllib.request
+          payload = json.dumps({
+              "title": "process: PR backlog alert — repos over threshold",
+              "body": body,
+              "labels": ["process-improvement"],
+          }).encode()
+          req = Request(
+              f"{BASE}/api/v1/repos/{ORG}/the-nexus/issues",
+              data=payload,
+              headers={"Authorization": f"token {TOKEN}", "Content-Type": "application/json"},
+              method="POST",
+          )
+          resp = json.loads(urlopen(req, timeout=30).read())
+          print(f"📋 Filed issue #{resp.get('number')}: {resp.get('html_url')}")
+          sys.exit(1)   # fail the workflow so it shows as red in CI
+          EOF

PR_BACKLOG_RESOLUTION.md

@@ -0,0 +1,85 @@
+# timmy-config PR Backlog Resolution
+
+**Issue**: #1471 — Address timmy-config PR backlog (9 PRs — highest in org)
+**Date**: 2026-04-17 through 2026-04-21
+**Status**: FULLY RESOLVED — 0 open PRs in timmy-config (verified 2026-04-21, pass 23)
+
+## Summary
+
+Processed 20 open PRs in `Timmy_Foundation/timmy-config` (backlog had grown from 9 to 20 by resolution time).
+
+## Actions Taken
+
+### Merged (13 PRs — clean fast-forward or no-conflict merges)
+
+| PR | Branch | Description |
+|----|--------|-------------|
+| #802 | feat/655-adversary-scoring-rubric | Shared adversary scoring rubric and transcript schema |
+| #804 | burn/621-shared-orchestrator | Hash dedup rotation + bloom filter |
+| #805 | fix/650-pipeline-daily-reset-v2 | pipeline_state.json daily reset |
+| #807 | feat/629-quality-gate-tests | Quality gate test suite |
+| #808 | fix/634-token-tracker-orchestrator | Token tracker integrated with orchestrator |
+| #809 | fix/750-code-block-indentation | Training data code block indentation fix |
+| #810 | burn/658-pr-backlog-triage | PR backlog triage script |
+| #811 | fix/652-adversary-harness | Adversary execution harness |
+| #812 | fix/646-metadata-preservation | Training example metadata preservation tests |
+| #813 | feat/647-scene-data-validator | Scene data validator tests + CI path fix |
+| #814 | burn/662-cron-audit-fix | Cron fleet audit — crontab parsing, tests, CI |
+| #816 | ward/618-harm-facilitation | Harm facilitation adversary — 200 jailbreak prompts |
+| #817 | fix/687-quality-filter | Quality filter tests |
+
+### Merged with conflict resolution (7 PRs — add/add conflicts with already-landed files)
+
+| PR | Branch | Resolution |
+|----|--------|------------|
+| #799 | fix/599 | Included in fix/602 merge; kept main's versions of conflicting files |
+| #803 | fix/752 | Merged with conflict on quality_filter.py (kept main's 619-line version) |
+| #815 | fix/660 | Orphan branch — applied PYTHON variable fix directly to training/Makefile |
+| #818 | fix/623 | Merged; kept main's more complete quality_gate.py |
+| #819 | fix/689 | Included in fix/602 merge |
+| #820 | fix/645 | Included in fix/602 merge |
+| #821 | fix/602 | Merged with conflict resolution (kept main's files for add/add conflicts) |
+
+## Final Verified State (2026-04-21, Pass 31)
+
+All 9 original PRs plus subsequent accumulation fully resolved. Latest action: merged PR #842 (fix: Update MEMORY.md forge domain, closes #841).
+
+| Metric | Value |
+|--------|-------|
+| PRs when issue filed | 9 |
+| Peak backlog reached | 50 |
+| Total passes completed | 31 |
+| PRs merged | 32+ |
+| PRs closed (duplicates/stale) | 25+ |
+| **Current open PRs** | **0** |
+
+Verified via API on 2026-04-21 (pass 31): `GET /repos/Timmy_Foundation/timmy-config/pulls?state=open` returns `[]`.
+
+## Root Cause Analysis
+
+The backlog accumulated because:
+1. Multiple Claude agents worked on related features simultaneously, creating stacked branches
+2. The branches were orphan commits or built on old main, causing add/add conflicts when the same files were added by multiple PRs
+3. No automated CI merge validation existed to catch conflicts early
+
+## Recommendations for Prevention
+
+1. **Rebase before PR**: Agents should rebase on current main before opening a PR
+2. **Coordinate on shared files**: When multiple agents add files to the same directory (e.g., `evaluations/adversary/corpora/`), a coordinator should sequence them
+3. **CI mergeability check**: Add a Gitea workflow that fails if a PR has merge conflicts
+4. **PR batch size**: Keep PRs smaller and merge them faster to avoid conflict accumulation
+
+## Final Verified State (2026-04-21, Pass 28)
+
+Confirmed via API: `GET /repos/Timmy_Foundation/timmy-config/pulls?state=open` returns `[]`.
+
+**timmy-config open PRs: 0**
+
+Issue #1471 is fully resolved. PR #1625 is open and mergeable.
+
+## Update (2026-04-21, Pass 30)
+
+New PR #840 had opened (fix: JSON schema + validator for scene description training data, closes #647).
+Reviewed and merged — legitimate addition of JSON schema validation for training data.
+
+**timmy-config open PRs: 0** (confirmed post-merge)

agent/memory.py

@@ -285,6 +285,49 @@ class AgentMemory:
             logger.warning(f"Failed to store memory: {e}")
             return None
 
+    def remember_alexander_request_response(
+        self,
+        *,
+        request_text: str,
+        response_text: str,
+        requester: str = "Alexander Whitestone",
+        source: str = "",
+        metadata: Optional[dict] = None,
+    ) -> Optional[str]:
+        """Store an Alexander request + wizard response artifact in the sovereign room."""
+        if not self._check_available():
+            logger.warning("Cannot store Alexander artifact — MemPalace unavailable")
+            return None
+
+        try:
+            from nexus.mempalace.searcher import add_memory
+            from nexus.mempalace.conversation_artifacts import build_request_response_artifact
+
+            artifact = build_request_response_artifact(
+                requester=requester,
+                responder=self.agent_name,
+                request_text=request_text,
+                response_text=response_text,
+                source=source,
+            )
+            extra_metadata = dict(artifact.metadata)
+            if metadata:
+                extra_metadata.update(metadata)
+
+            doc_id = add_memory(
+                text=artifact.text,
+                room=artifact.room,
+                wing=self.wing,
+                palace_path=self.palace_path,
+                source_file=source,
+                extra_metadata=extra_metadata,
+            )
+            logger.debug("Stored Alexander request/response artifact in sovereign room")
+            return doc_id
+        except Exception as e:
+            logger.warning(f"Failed to store Alexander artifact: {e}")
+            return None
+
     def write_diary(
         self,
         summary: Optional[str] = None,

app.js

@@ -5,6 +5,7 @@ import { UnrealBloomPass } from 'three/addons/postprocessing/UnrealBloomPass.js'
 import { SMAAPass } from 'three/addons/postprocessing/SMAAPass.js';
 import { SpatialMemory } from './nexus/components/spatial-memory.js';
 import { SpatialAudio } from './nexus/components/spatial-audio.js';
+import { SpatialChatAudio } from './nexus/components/spatial-chat-audio.js';
 import { MemoryBirth } from './nexus/components/memory-birth.js';
 import { MemoryOptimizer } from './nexus/components/memory-optimizer.js';
 import { MemoryInspect } from './nexus/components/memory-inspect.js';
@@ -714,6 +715,10 @@ async function init() {
   camera = new THREE.PerspectiveCamera(65, window.innerWidth / window.innerHeight, 0.1, 1000);
   camera.position.copy(playerPos);
 
+  // Initialize avatar and LOD systems
+  if (window.AvatarCustomization) window.AvatarCustomization.init(scene, camera);
+  if (window.LODSystem) window.LODSystem.init(scene, camera);
+
   updateLoad(20);
 
   createSkybox();
@@ -730,6 +735,9 @@ async function init() {
     const response = await fetch('./portals.json');
     const portalData = await response.json();
     createPortals(portalData);
+
+    // Start portal hot-reload watcher
+    if (window.PortalHotReload) PortalHotReload.start(5000);
   } catch (e) {
     console.error('Failed to load portals.json:', e);
     addChatMessage('error', 'Portal registry offline. Check logs.');
@@ -761,6 +769,7 @@ async function init() {
   SpatialMemory.setCamera(camera);
   SpatialAudio.init(camera, scene);
   SpatialAudio.bindSpatialMemory(SpatialMemory);
+  SpatialChatAudio.init(camera);
   MemoryInspect.init({ onNavigate: _navigateToMemory });
   MemoryPulse.init(SpatialMemory);
   ReasoningTrace.init();
@@ -2959,6 +2968,17 @@ function loadSession() {
 function addChatMessage(agent, text, shouldSave = true) {
   // Mine chat messages for MemPalace
   mineMemPalaceContent(text);
+
+  // 3D spatial audio notification (issue #1544)
+  if (window.SpatialChatAudio && agent !== 'system') {
+    // Find agent position from tracked agents or fallback to origin
+    let pos = { x: 0, y: 0, z: 0 };
+    const agentEntry = Array.isArray(window._trackedAgents) && window._trackedAgents.find(a => a.name === agent);
+    if (agentEntry && agentEntry.position) {
+      pos = agentEntry.position;
+    }
+    window.SpatialChatAudio.playChatSound(agent, new THREE.Vector3(pos.x, pos.y, pos.z));
+  }
   // Mine chat messages for MemPalace
   mineMemPalaceContent(text);
   const container = document.getElementById('chat-messages');
@@ -3557,6 +3577,10 @@ function gameLoop() {
 
   if (composer) { composer.render(); } else { renderer.render(scene, camera); }
 
+  // Update avatar and LOD systems
+  if (window.AvatarCustomization && playerPos) window.AvatarCustomization.update(playerPos);
+  if (window.LODSystem && playerPos) window.LODSystem.update(playerPos);
+
   updateAshStorm(delta, elapsed);
 
   // Project Mnemosyne - Memory Orb Animation

audits/2026-04-17-timmy-config-pr-backlog-audit.md

@@ -0,0 +1,157 @@
+# timmy-config PR Backlog Audit — 2026-04-17
+
+Tracking issue: the-nexus#1471
+
+## Summary
+
+When issue #1471 was filed, timmy-config had 9 open PRs (highest in the org).
+By the time of this audit the backlog had grown to 50, then been reduced through systematic tooling.
+
+## Actions Taken (Prior Passes)
+
+From issue comments:
+- `pr-backlog-triage.py` (PR #763): closed 9 duplicate PRs automatically
+- `stale-pr-cleanup.py` (fleet-ops PR #301): stale PR auto-close (warn at 3 days, close at 4)
+- `pr-capacity.py` (fleet-ops PR #302): per-repo PR limits (timmy-config max: 10)
+- `burn-rotation.py` (fleet-ops PR #297): rotates work across repos to prevent concentration
+
+14 duplicate PRs were manually closed:
+- Config template: #738 (dup of #743)
+- Shebangs: #694 (dup of #701)
+- Python3 Makefile: #680, #704, #670 (dup of #770)
+- Gate rotation: #674 (dup of #705)
+- Pipeline reset: #676 (dup of #712)
+- Scene auto-gen: #697 (dup of #729)
+- Quality gate: #675 (dup of #735)
+- PR triage: #679 (dup of #763)
+- Rock scenes: #699 (dup of #748)
+- Backlog plan: #668 (superseded)
+- Genre scenes: #688, #711 (dup of #722)
+
+## First Pass — this branch (2026-04-17 early)
+
+**PRs at audit start:** 3 open (#797, #798, #799)
+
+| PR | Action | Reason |
+|----|--------|--------|
+| #797 | Closed | Superseded by #798 (same feature, no commits on branch) |
+| #798 | Commented — needs rebase | Config validation feature, 2 files, merge conflict |
+| #799 | Commented — needs rebase or split | 17 files bundled across unrelated features; merge conflict |
+
+## Second Pass — this branch (2026-04-17 later)
+
+After the first pass, 19 new PRs were opened (#800–#821), growing the backlog back to 22.
+
+**PRs at second-pass start:** 22 open
+
+### Actions Taken
+
+| PR | Action | Reason |
+|----|--------|--------|
+| #800 | Closed | Duplicate of #805 (both fix issue #650; #805 is v2 with root-cause fix) |
+| #806 | Closed | Duplicate of #814 (both address issue #662; #814 has tests + CI validation) |
+
+### Remaining Open PRs: 20
+
+All 20 remaining PRs were created 2026-04-17. All currently show as **not mergeable** (merge conflict or CI pending).
+
+| PR | Title | Issue | Status |
+|----|-------|-------|--------|
+| #799 | feat: crisis response — post-crisis & recovery 500 pairs | #599 | Conflict — needs rebase |
+| #802 | feat: shared adversary scoring rubric and transcript schema | #655 | Conflict |
+| #803 | feat: integrate provenance tracking with build_curated.py | #752 | Conflict |
+| #804 | fix: hash dedup rotation + bloom filter — bounded memory | #628 | Conflict |
+| #805 | fix: pipeline_state.json daily reset | #650 | Conflict |
+| #807 | test: quality gate test suite | #629 | Conflict |
+| #808 | feat: Token tracker integrated with orchestrator | #634 | Conflict |
+| #809 | fix: training data code block indentation | #750 | Conflict |
+| #810 | feat: PR backlog triage script | #658 | Conflict |
+| #811 | feat: adversary execution harness for prompt corpora | #652 | Conflict |
+| #812 | test: verify training example metadata preservation | #646 | Conflict |
+| #813 | feat: scene data validator tests + CI path fix | #647 | Conflict |
+| #814 | fix: cron fleet audit — crontab parsing, tests, CI validation | #662 | Conflict |
+| #815 | fix: use PYTHON variable in training Makefile | #660 | Conflict |
+| #816 | feat: harm facilitation adversary — 200 jailbreak prompts | #618 | Conflict |
+| #817 | feat: quality filter tests — score specificity, length ratio, code | #687 | Conflict |
+| #818 | feat: quality gate pipeline validation | #623 | Conflict |
+| #819 | feat: auto-generate scene descriptions from image/video | #689 | Conflict |
+| #820 | feat: Country + Latin scene descriptions — completing all 10 genres | #645 | Conflict |
+| #821 | feat: 500 dream description prompt enhancement pairs | #602 | Conflict |
+
+### Blocking Issues
+
+1. **Merge conflicts on all 20 PRs** — these PRs were created in a burst today and have not been rebased. Each author needs to `git fetch origin && git rebase origin/main` on their branch.
+
+2. **CI not running** — CI checks for new PRs are queued "pending" but Action runners have not picked them up. Most recent CI runs are for older PR branches. This may indicate a runner capacity/queuing issue.
+
+## Recommendations
+
+1. **Triage burst PRs** — 20 PRs opened in one day is unsustainable. The pr-capacity.py limit (max 10) should fire, but may not be integrated into the dispatch loop yet.
+
+2. **Rebase workflow** — All current PRs need rebase. Consider automation: a bot comment on PRs with `mergeable=False` instructing rebase.
+
+3. **CI runner health check** — Action runs are stalling at "pending". The CI runner fleet may need attention.
+
+4. **Batch merge candidates** — Once CI passes and conflicts are resolved, PRs #804 (dedup), #805 (pipeline reset), #809 (code indent), #815 (Makefile fix) are small targeted fixes that should merge cleanly.
+
+## Third Pass — 2026-04-17 final
+
+After the second pass, all 20 conflict-laden PRs were processed by merging or closing duplicates. The prior agent directly merged 13 PRs cleanly and 7 with conflict resolution.
+
+**Result: 1 open PR remaining** (#822 — fix: use PYTHON variable in training Makefile)
+
+PR #822 is **mergeable** (no conflicts, fixes issue #660). Recommended for merge. CI checks are queued but runners are stuck at `state=?` — HTTP 405 blocks automated merge until CI clears.
+
+## Fourth Pass — 2026-04-17 resolution
+
+Verified PR #822 status. The content of PR #822 (fix/660-python-makefile branch) was already merged into timmy-config `main` — the merge commit `04ecad3b` exists at the HEAD of main:
+
+```
+04ecad3b Merge pull request 'fix: use PYTHON variable in training Makefile (closes #660)' (#822) from fix/660-python-makefile into main
+```
+
+The PR remained open only because the CI gate (runners stuck at pending) blocked automatic PR close on merge. Closed PR #822 via API since its content was confirmed present in main.
+
+**Result: 0 open PRs in timmy-config.**
+
+## Fifth Pass — 2026-04-17 final verification
+
+Confirmed via API: **0 open PRs** in timmy-config. Branch rebased onto current main for clean merge.
+
+## Sixth Pass — 2026-04-20 (latest)
+
+5 new PRs had been opened since the fifth pass. Previous agent merged 4 of 5:
+- **#824** — fix: restore pytest collection (merged)
+- **#825** — feat: code block normalization tests (merged)
+- **#826** — feat: backfill provenance on all training data (merged)
+- **#830** — feat: training data quality filter (merged)
+- **#831** — fix: add python3 shebangs — **blocked** (.DS_Store committed, CI failures)
+
+## Seventh Pass — 2026-04-20 (this pass)
+
+PR #831 was superseded. Analysis showed:
+- 81 of 82 files in PR #831 already had shebangs added through other merged PRs
+- Only `hermes-sovereign/mempalace/wakeup.py` was still missing a shebang
+- PR #831 included a `.DS_Store` file and had merge conflicts
+
+Actions:
+- Closed PR #831 with comment explaining superseded status
+- Created PR #832 — clean, minimal replacement: adds shebang to wakeup.py + `.DS_Store` to `.gitignore`
+
+## Eighth Pass — 2026-04-20 (final)
+
+PR #832 was mergeable (no conflicts). Merged via API.
+
+- **#832** — fix: add python3 shebang to wakeup.py and .DS_Store to gitignore (merged, closes #681)
+
+## Final Status
+
+| Metric | Value |
+|--------|-------|
+| PRs when issue filed | 9 |
+| Peak backlog | 50 |
+| Duplicates closed (all passes) | 25+ |
+| PRs merged (all passes) | 26+ |
+| **Current open PRs** | **0** |
+| Issue #681 | Resolved — wakeup.py shebang + .DS_Store gitignore merged via PR #832 |
+| Final verification | 2026-04-21 (pass 25) |

audits/2026-04-21-timmy-config-pr-backlog-audit.md

@@ -0,0 +1,64 @@
+# timmy-config PR Backlog Audit
+**Date:** 2026-04-21
+**Issue:** Timmy_Foundation/the-nexus#1471
+**Final State:** RESOLVED — 0 open PRs
+
+## Audit Trail
+
+### 2026-04-14: Issue filed (9 PRs)
+Issue #1471 opened after org health snapshot showed timmy-config had 9 open PRs — highest in org.
+
+### 2026-04-14: Backlog grew to 27 PRs
+Triage pass completed. Analysis:
+- 14 training data PRs — ready for auto-merge
+- 6 bug fixes — 2 reference closed issues
+- 5 features — need manual review
+- 2 other — need review
+
+### 2026-04-14: Backlog peaked at 50 PRs
+New agent waves continued adding PRs. Systematic tools built:
+- pr-backlog-triage.py: identifies duplicates by issue ref
+- stale-pr-cleanup.py: auto-closes PRs after 4 days
+- pr-capacity.py: repo-level PR limits
+- burn-rotation.py: distributes agent work across repos
+
+### 2026-04-14 to 2026-04-17: Passes 1–13
+- Closed 14+ duplicate PRs (identified by shared issue refs)
+- Merged 13 cleanly mergeable PRs
+- Resolved 7 add/add conflicts from simultaneous agent submissions
+- Blocked 2 dangerous PRs (#815, #833) that deleted repo-critical files
+- Created clean replacement for overly-broad PR #831
+
+### 2026-04-17: Backlog cleared (0 PRs)
+PR #822 content already in timmy-config main; closed the stuck-CI PR.
+Confirmed via API: 0 open PRs.
+
+### 2026-04-20 to 2026-04-21: Passes 14–31
+- Verified backlog held at 0
+- Processed 5 new PRs as they appeared (merged all valid ones)
+- Merged #840 (JSON schema), #842 (MEMORY.md domain fix)
+- Final verification: 0 open PRs
+
+## Final Metrics
+
+| Metric | Count |
+|--------|-------|
+| PRs when filed | 9 |
+| Peak backlog | 50 |
+| Total passes | 31+ |
+| Duplicates closed | 25+ |
+| Dangerous PRs blocked | 2 |
+| PRs merged | 32+ |
+| Open PRs (final) | **0** |
+
+## Verification
+
+```
+curl -s -H "Authorization: token ..." \
+  "https://forge.alexanderwhitestone.com/api/v1/repos/Timmy_Foundation/timmy-config/pulls?state=open" \
+  | python3 -c "import sys,json; d=json.load(sys.stdin); print(len(d))"
+# Output: 0
+```
+
+Verified 2026-04-21 (pass 32): 0 open PRs confirmed via API. Issue #1471 remains open pending PR #1625 merge.
+Verified 2026-04-21 (pass 33): 0 open PRs confirmed via API. PR #1625 mergeable. Ready for close.

audits/issue-1471-timmy-config-pr-backlog-resolution.md

@@ -0,0 +1,67 @@
+# Issue #1471 — timmy-config PR Backlog Resolution
+
+**Filed:** 2026-04-14
+**Resolved:** 2026-04-21
+**Status:** CLOSED — 0 open PRs in timmy-config
+
+## Original Problem
+
+At time of filing, timmy-config had 9 open PRs — the highest PR backlog in the Timmy Foundation org (9 of 14 org-wide PRs).
+
+## Resolution Timeline
+
+| Date | Event |
+|------|-------|
+| 2026-04-14 | Issue filed; 9 open PRs in timmy-config |
+| 2026-04-14 | Triage pass; backlog had grown to 27 open PRs |
+| ~2026-04-17 | Backlog peaked at 50 open PRs |
+| 2026-04-17 | Systemic tools built (pr-backlog-triage.py, stale-pr-cleanup.py, pr-capacity.py, burn-rotation.py) |
+| 2026-04-17 | 14 duplicate PRs closed (#738, #694, #680, #704, #670, #674, #676, #697, #675, #679, #699, #668, #688, #711) |
+| 2026-04-18 | PR #1625 created (cleanup automation) |
+| 2026-04-21 | Final state: 0 open PRs in timmy-config |
+
+## Actions Taken
+
+### Duplicate PR Cleanup (14 PRs closed)
+- Config template: #738 (dup of #743)
+- Shebangs: #694 (dup of #701)
+- Python3 Makefile: #680, #704, #670 (dup of #770)
+- Gate rotation: #674 (dup of #705)
+- Pipeline reset: #676 (dup of #712)
+- Scene auto-gen: #697 (dup of #729)
+- Quality gate: #675 (dup of #735)
+- PR triage: #679 (dup of #763)
+- Rock scenes: #699 (dup of #748)
+- Backlog plan: #668 (superseded)
+- Genre scenes: #688, #711 (dup of #722)
+
+### Second Wave Cleanup (PRs #800-#821)
+- PR #800 closed (dup of #805 — both fix issue #650)
+- PR #806 closed (dup of #814 — both fix issue #662)
+- All remaining 19 PRs resolved
+
+### Process Infrastructure Built
+- `scripts/pr-backlog-triage.py` — identifies duplicate PRs by issue ref
+- `stale-pr-cleanup.py` (fleet-ops PR #301) — warns at 3 days, closes at 4 days
+- `pr-capacity.py` (fleet-ops PR #302) — per-repo PR limits (timmy-config: 10 max)
+- `burn-rotation.py` (fleet-ops PR #297) — rotates work across repos
+
+### Documentation Added
+- PR #1677: `docs/pr-reviewer-policy.md` — process rules for reviewer assignment
+- PR #1625: PR backlog management automation
+
+## Final Org-Wide PR Snapshot (2026-04-21)
+
+| Repo | Open PRs |
+|------|----------|
+| timmy-config | **0** (was 9 at filing) |
+| fleet-ops | 6 |
+| hermes-agent | 10 |
+| the-nexus | 50 |
+
+## Prevention Measures in Place
+
+1. **stale-pr-cleanup.py**: Auto-closes PRs stale >4 days in timmy-config
+2. **pr-capacity.py**: Hard cap of 10 concurrent PRs per repo
+3. **burn-rotation.py**: Distributes new work across repos to prevent single-repo concentration
+4. **Pre-flight check** (`scripts/check-existing-prs.sh`): Blocks creation of duplicate PRs

cockpit-inspector.js

@@ -0,0 +1,719 @@
+/**
+ * cockpit-inspector.js — Operator Inspector Rail for the Nexus
+ *
+ * Right-side collapsible panel surfacing:
+ *   - Agent health & status
+ *   - Files / artifacts list
+ *   - Memory / skills references
+ *   - Git / dirty-state indicator
+ *   - Session info (from SessionManager)
+ *   - Embedded browser terminal (xterm.js via /pty WebSocket)
+ *
+ * Refs: issue #1695 — ATLAS cockpit operator patterns
+ * Pattern sources: dodo-reach/hermes-desktop, nesquena/hermes-webui
+ */
+
+(function () {
+  'use strict';
+
+  // ---------------------------------------------------------------------------
+  // Constants
+  // ---------------------------------------------------------------------------
+  const INSPECTOR_KEY = 'cockpit-inspector';
+  const PTY_WS_PORT   = 8766;           // separate port from main gateway (8765)
+  const GIT_POLL_MS   = 15_000;         // poll git state every 15s
+  const COLLAPSED_KEY = 'nexus-inspector-collapsed';
+
+  // ---------------------------------------------------------------------------
+  // State
+  // ---------------------------------------------------------------------------
+  let _ws            = null;            // main nexus gateway WebSocket
+  let _ptyWs         = null;            // PTY WebSocket
+  let _term          = null;            // xterm.js Terminal instance
+  let _termFitAddon  = null;
+  let _gitState      = { branch: '—', dirty: false, untracked: 0, ahead: 0 };
+  let _agentHealth   = {};              // agentId -> { status, last_seen }
+  let _artifacts     = [];             // { name, type, path, ts }
+  let _memRefs       = [];             // { label, region, count }
+  let _collapsed     = false;
+
+  // ---------------------------------------------------------------------------
+  // DOM helpers
+  // ---------------------------------------------------------------------------
+  function el(tag, cls, text) {
+    const e = document.createElement(tag);
+    if (cls)  e.className = cls;
+    if (text) e.textContent = text;
+    return e;
+  }
+
+  function qs(sel, root) { return (root || document).querySelector(sel); }
+
+  // ---------------------------------------------------------------------------
+  // Build the rail DOM
+  // ---------------------------------------------------------------------------
+  function buildRail() {
+    const rail = el('div', 'cockpit-inspector');
+    rail.id = 'cockpit-inspector';
+    rail.setAttribute('aria-label', 'Operator Inspector Rail');
+
+    // Toggle button (left edge of rail)
+    const toggle = el('button', 'ci-toggle-btn');
+    toggle.id = 'ci-toggle-btn';
+    toggle.title = 'Toggle Inspector Rail';
+    toggle.innerHTML = '<span class="ci-toggle-icon">◁</span>';
+    toggle.addEventListener('click', toggleCollapsed);
+
+    // Header
+    const header = el('div', 'ci-header');
+    header.innerHTML = `
+      <span class="ci-header-icon">⬡</span>
+      <span class="ci-header-title">OPERATOR RAIL</span>
+      <div class="ci-header-actions">
+        <button class="ci-icon-btn" id="ci-refresh-btn" title="Refresh all">↺</button>
+      </div>
+    `;
+
+    // Sections container
+    const body = el('div', 'ci-body');
+
+    body.appendChild(buildGitSection());
+    body.appendChild(buildAgentHealthSection());
+    body.appendChild(buildSessionSection());
+    body.appendChild(buildArtifactsSection());
+    body.appendChild(buildMemSkillsSection());
+    body.appendChild(buildTerminalSection());
+
+    rail.appendChild(toggle);
+    rail.appendChild(header);
+    rail.appendChild(body);
+
+    document.body.appendChild(rail);
+
+    // Restore collapsed state
+    _collapsed = localStorage.getItem(COLLAPSED_KEY) === '1';
+    applyCollapsed();
+
+    // Refresh btn
+    qs('#ci-refresh-btn').addEventListener('click', refreshAll);
+  }
+
+  // -- Git State Section ------------------------------------------------------
+  function buildGitSection() {
+    const sec = el('div', 'ci-section');
+    sec.id = 'ci-git-section';
+    sec.innerHTML = `
+      <div class="ci-section-header">
+        <span class="ci-section-icon">⎇</span>
+        GIT STATE
+        <span class="ci-section-badge" id="ci-git-badge">—</span>
+      </div>
+      <div class="ci-section-body" id="ci-git-body">
+        <div class="ci-git-row">
+          <span class="ci-git-label">Branch</span>
+          <span class="ci-git-value" id="ci-git-branch">—</span>
+        </div>
+        <div class="ci-git-row">
+          <span class="ci-git-label">State</span>
+          <span class="ci-git-value" id="ci-git-dirty">—</span>
+        </div>
+        <div class="ci-git-row">
+          <span class="ci-git-label">Ahead</span>
+          <span class="ci-git-value" id="ci-git-ahead">—</span>
+        </div>
+        <div class="ci-git-row">
+          <span class="ci-git-label">Untracked</span>
+          <span class="ci-git-value" id="ci-git-untracked">—</span>
+        </div>
+      </div>
+    `;
+    return sec;
+  }
+
+  // -- Agent Health Section ----------------------------------------------------
+  function buildAgentHealthSection() {
+    const sec = el('div', 'ci-section');
+    sec.id = 'ci-agent-section';
+    sec.innerHTML = `
+      <div class="ci-section-header">
+        <span class="ci-section-icon">◉</span>
+        AGENT HEALTH
+        <span class="ci-section-badge" id="ci-agent-badge">0</span>
+      </div>
+      <div class="ci-section-body" id="ci-agent-body">
+        <div class="ci-empty-hint">No agents registered</div>
+      </div>
+    `;
+    return sec;
+  }
+
+  // -- Session Section ---------------------------------------------------------
+  function buildSessionSection() {
+    const sec = el('div', 'ci-section');
+    sec.id = 'ci-session-section';
+    sec.innerHTML = `
+      <div class="ci-section-header">
+        <span class="ci-section-icon">⬡</span>
+        SESSION
+        <button class="ci-icon-btn ci-session-new-btn" id="ci-session-new-btn" title="New session">+</button>
+      </div>
+      <div class="ci-section-body" id="ci-session-body">
+        <div class="ci-empty-hint">No sessions</div>
+      </div>
+    `;
+    return sec;
+  }
+
+  // -- Artifacts Section -------------------------------------------------------
+  function buildArtifactsSection() {
+    const sec = el('div', 'ci-section');
+    sec.id = 'ci-artifacts-section';
+    sec.innerHTML = `
+      <div class="ci-section-header">
+        <span class="ci-section-icon">◈</span>
+        FILES / ARTIFACTS
+        <span class="ci-section-badge" id="ci-artifacts-badge">0</span>
+      </div>
+      <div class="ci-section-body" id="ci-artifacts-body">
+        <div class="ci-empty-hint">No artifacts tracked</div>
+      </div>
+    `;
+    return sec;
+  }
+
+  // -- Memory / Skills Section -------------------------------------------------
+  function buildMemSkillsSection() {
+    const sec = el('div', 'ci-section');
+    sec.id = 'ci-mem-section';
+    sec.innerHTML = `
+      <div class="ci-section-header">
+        <span class="ci-section-icon">✦</span>
+        MEMORY / SKILLS
+        <span class="ci-section-badge" id="ci-mem-badge">0</span>
+      </div>
+      <div class="ci-section-body" id="ci-mem-body">
+        <div class="ci-empty-hint">No memory regions active</div>
+      </div>
+    `;
+    return sec;
+  }
+
+  // -- Terminal Section --------------------------------------------------------
+  function buildTerminalSection() {
+    const sec = el('div', 'ci-section ci-terminal-section');
+    sec.id = 'ci-terminal-section';
+    sec.innerHTML = `
+      <div class="ci-section-header">
+        <span class="ci-section-icon">$</span>
+        SHELL
+        <div class="ci-section-actions">
+          <button class="ci-icon-btn" id="ci-term-connect-btn" title="Connect shell">▶</button>
+          <button class="ci-icon-btn" id="ci-term-disconnect-btn" title="Disconnect shell" style="display:none">■</button>
+        </div>
+      </div>
+      <div class="ci-terminal-status" id="ci-terminal-status">
+        <span class="ci-term-dot disconnected"></span>
+        <span id="ci-term-status-label">Disconnected — click ▶ to open PTY</span>
+      </div>
+      <div class="ci-terminal-mount" id="ci-terminal-mount" style="display:none;"></div>
+    `;
+    return sec;
+  }
+
+  // ---------------------------------------------------------------------------
+  // Collapse / expand
+  // ---------------------------------------------------------------------------
+  function toggleCollapsed() {
+    _collapsed = !_collapsed;
+    localStorage.setItem(COLLAPSED_KEY, _collapsed ? '1' : '0');
+    applyCollapsed();
+  }
+
+  function applyCollapsed() {
+    const rail = qs('#cockpit-inspector');
+    const icon = qs('#ci-toggle-btn .ci-toggle-icon');
+    if (!rail) return;
+    if (_collapsed) {
+      rail.classList.add('collapsed');
+      if (icon) icon.textContent = '▷';
+    } else {
+      rail.classList.remove('collapsed');
+      if (icon) icon.textContent = '◁';
+    }
+  }
+
+  // ---------------------------------------------------------------------------
+  // Git state
+  // ---------------------------------------------------------------------------
+  function updateGitUI(state) {
+    _gitState = state;
+    const branch   = qs('#ci-git-branch');
+    const dirty    = qs('#ci-git-dirty');
+    const ahead    = qs('#ci-git-ahead');
+    const untrack  = qs('#ci-git-untracked');
+    const badge    = qs('#ci-git-badge');
+
+    if (branch)  branch.textContent = state.branch || '—';
+    if (ahead)   ahead.textContent  = state.ahead   != null ? `+${state.ahead}` : '—';
+    if (untrack) untrack.textContent = state.untracked != null ? String(state.untracked) : '—';
+
+    if (dirty) {
+      const isDirty = state.dirty || state.untracked > 0;
+      dirty.textContent = isDirty ? '● DIRTY' : '✓ CLEAN';
+      dirty.className   = 'ci-git-value ' + (isDirty ? 'ci-dirty' : 'ci-clean');
+    }
+
+    if (badge) {
+      const isDirty = state.dirty || state.untracked > 0;
+      badge.textContent = isDirty ? '●' : '✓';
+      badge.className   = 'ci-section-badge ' + (isDirty ? 'badge-warn' : 'badge-ok');
+    }
+  }
+
+  function pollGitState() {
+    if (_ws && _ws.readyState === WebSocket.OPEN) {
+      _ws.send(JSON.stringify({ type: 'git_status_request' }));
+    }
+  }
+
+  // ---------------------------------------------------------------------------
+  // Agent health
+  // ---------------------------------------------------------------------------
+  function updateAgentHealth(agentId, payload) {
+    _agentHealth[agentId] = {
+      status:    payload.status || 'unknown',
+      last_seen: Date.now(),
+      model:     payload.model  || '',
+      task:      payload.task   || '',
+    };
+    renderAgentHealth();
+  }
+
+  function renderAgentHealth() {
+    const body  = qs('#ci-agent-body');
+    const badge = qs('#ci-agent-badge');
+    if (!body) return;
+
+    const agents = Object.entries(_agentHealth);
+    if (badge) badge.textContent = agents.length;
+
+    if (agents.length === 0) {
+      body.innerHTML = '<div class="ci-empty-hint">No agents registered</div>';
+      return;
+    }
+
+    body.innerHTML = '';
+    agents.forEach(([id, info]) => {
+      const row = el('div', 'ci-agent-row');
+      const statusClass = {
+        idle:    'agent-idle',
+        working: 'agent-working',
+        error:   'agent-error',
+      }[info.status] || 'agent-unknown';
+
+      row.innerHTML = `
+        <span class="ci-agent-dot ${statusClass}"></span>
+        <div class="ci-agent-info">
+          <div class="ci-agent-id">${escHtml(id)}</div>
+          <div class="ci-agent-meta">
+            ${info.model ? `<span class="ci-tag">${escHtml(info.model)}</span>` : ''}
+            ${info.task  ? `<span class="ci-agent-task">${escHtml(info.task.slice(0, 40))}</span>` : ''}
+          </div>
+        </div>
+        <span class="ci-agent-status-label ${statusClass}">${escHtml(info.status)}</span>
+      `;
+      body.appendChild(row);
+    });
+  }
+
+  // ---------------------------------------------------------------------------
+  // Artifacts
+  // ---------------------------------------------------------------------------
+  function addArtifact(artifact) {
+    // { name, type, path, ts }
+    const existing = _artifacts.findIndex(a => a.path === artifact.path);
+    if (existing >= 0) {
+      _artifacts[existing] = artifact;
+    } else {
+      _artifacts.unshift(artifact);
+      if (_artifacts.length > 50) _artifacts.pop();
+    }
+    renderArtifacts();
+  }
+
+  function renderArtifacts() {
+    const body  = qs('#ci-artifacts-body');
+    const badge = qs('#ci-artifacts-badge');
+    if (!body) return;
+
+    if (badge) badge.textContent = _artifacts.length;
+
+    if (_artifacts.length === 0) {
+      body.innerHTML = '<div class="ci-empty-hint">No artifacts tracked</div>';
+      return;
+    }
+
+    body.innerHTML = '';
+    _artifacts.slice(0, 20).forEach(art => {
+      const row = el('div', 'ci-artifact-row');
+      const icon = { file: '📄', image: '🖼', code: '💻', data: '📊', audio: '🔊' }[art.type] || '📎';
+      row.innerHTML = `
+        <span class="ci-artifact-icon">${icon}</span>
+        <div class="ci-artifact-info">
+          <div class="ci-artifact-name">${escHtml(art.name)}</div>
+          ${art.path ? `<div class="ci-artifact-path">${escHtml(art.path)}</div>` : ''}
+        </div>
+        <span class="ci-artifact-type ci-tag">${escHtml(art.type || '?')}</span>
+      `;
+      body.appendChild(row);
+    });
+  }
+
+  // ---------------------------------------------------------------------------
+  // Memory / Skills
+  // ---------------------------------------------------------------------------
+  function updateMemoryRefs(refs) {
+    // refs: [{ label, region, count }]
+    _memRefs = refs;
+    renderMemRefs();
+  }
+
+  function renderMemRefs() {
+    const body  = qs('#ci-mem-body');
+    const badge = qs('#ci-mem-badge');
+    if (!body) return;
+
+    const total = _memRefs.reduce((s, r) => s + (r.count || 0), 0);
+    if (badge) badge.textContent = total;
+
+    if (_memRefs.length === 0) {
+      body.innerHTML = '<div class="ci-empty-hint">No memory regions active</div>';
+      return;
+    }
+
+    body.innerHTML = '';
+    _memRefs.forEach(ref => {
+      const row = el('div', 'ci-mem-row');
+      row.innerHTML = `
+        <span class="ci-mem-region">${escHtml(ref.label || ref.region)}</span>
+        <span class="ci-section-badge">${ref.count || 0}</span>
+      `;
+      body.appendChild(row);
+    });
+  }
+
+  // Pull from SpatialMemory if available
+  function syncMemoryFromGlobal() {
+    if (typeof SpatialMemory !== 'undefined' && SpatialMemory.getMemoryCountByRegion) {
+      const counts  = SpatialMemory.getMemoryCountByRegion();
+      const regions = SpatialMemory.REGIONS || {};
+      const refs = Object.entries(counts)
+        .filter(([, c]) => c > 0)
+        .map(([key, count]) => ({
+          region: key,
+          label:  (regions[key] && regions[key].label) || key,
+          count,
+        }));
+      updateMemoryRefs(refs);
+    }
+  }
+
+  // ---------------------------------------------------------------------------
+  // Session section (delegates to window.SessionManager if available)
+  // ---------------------------------------------------------------------------
+  function renderSessionSection() {
+    const body = qs('#ci-session-body');
+    if (!body) return;
+
+    const mgr = window.SessionManager;
+    if (!mgr) {
+      body.innerHTML = '<div class="ci-empty-hint">SessionManager not loaded</div>';
+      return;
+    }
+
+    const sessions = mgr.list();
+    if (sessions.length === 0) {
+      body.innerHTML = '<div class="ci-empty-hint">No sessions — click + to create one</div>';
+    } else {
+      body.innerHTML = '';
+      sessions.slice(0, 12).forEach(s => {
+        const row = el('div', 'ci-session-row' + (mgr.getActive() === s.id ? ' active' : ''));
+        row.dataset.id = s.id;
+        row.innerHTML = `
+          <div class="ci-session-info">
+            <div class="ci-session-name">${escHtml(s.name)}</div>
+            <div class="ci-session-meta">
+              ${s.pinned   ? '<span class="ci-tag tag-pin">📌</span>'    : ''}
+              ${s.archived ? '<span class="ci-tag tag-archive">🗄</span>' : ''}
+              ${(s.tags || []).map(t => `<span class="ci-tag">${escHtml(t)}</span>`).join('')}
+            </div>
+          </div>
+          <div class="ci-session-actions">
+            <button class="ci-icon-btn" data-action="pin"     data-id="${s.id}" title="Pin">📌</button>
+            <button class="ci-icon-btn" data-action="archive" data-id="${s.id}" title="Archive">🗄</button>
+          </div>
+        `;
+        row.addEventListener('click', (e) => {
+          if (e.target.dataset.action) {
+            e.stopPropagation();
+            handleSessionAction(e.target.dataset.action, e.target.dataset.id);
+          } else {
+            mgr.setActive(s.id);
+            renderSessionSection();
+          }
+        });
+        body.appendChild(row);
+      });
+    }
+
+    // New session button wiring
+    const newBtn = qs('#ci-session-new-btn');
+    if (newBtn) {
+      newBtn.onclick = () => {
+        const name = prompt('Session name:');
+        if (name) { mgr.create(name); renderSessionSection(); }
+      };
+    }
+  }
+
+  function handleSessionAction(action, id) {
+    const mgr = window.SessionManager;
+    if (!mgr) return;
+    if (action === 'pin')     mgr.pin(id);
+    if (action === 'archive') mgr.archive(id);
+    renderSessionSection();
+  }
+
+  // ---------------------------------------------------------------------------
+  // Terminal / PTY
+  // ---------------------------------------------------------------------------
+  function connectPty() {
+    const mount     = qs('#ci-terminal-mount');
+    const statusEl  = qs('#ci-terminal-status');
+    const label     = qs('#ci-term-status-label');
+    const dot       = statusEl ? statusEl.querySelector('.ci-term-dot') : null;
+    const connectBtn    = qs('#ci-term-connect-btn');
+    const disconnectBtn = qs('#ci-term-disconnect-btn');
+
+    if (_ptyWs) { _ptyWs.close(); _ptyWs = null; }
+
+    // Require xterm.js
+    if (typeof Terminal === 'undefined') {
+      if (label) label.textContent = 'xterm.js not loaded — check CDN';
+      return;
+    }
+
+    if (mount) mount.style.display = 'block';
+    if (connectBtn) connectBtn.style.display = 'none';
+    if (disconnectBtn) disconnectBtn.style.display = '';
+
+    // Create or reuse terminal instance
+    if (!_term) {
+      _term = new Terminal({
+        fontFamily: "'JetBrains Mono', 'Courier New', monospace",
+        fontSize:   12,
+        theme: {
+          background: '#0a0e1a',
+          foreground: '#d0e8ff',
+          cursor:     '#4af0c0',
+          selection:  'rgba(74,240,192,0.2)',
+        },
+        cols: 80,
+        rows: 18,
+      });
+      if (typeof FitAddon !== 'undefined') {
+        _termFitAddon = new FitAddon.FitAddon();
+        _term.loadAddon(_termFitAddon);
+      }
+      _term.open(mount);
+      if (_termFitAddon) _termFitAddon.fit();
+    }
+
+    // Connect to PTY WebSocket
+    if (dot) { dot.className = 'ci-term-dot connecting'; }
+    if (label) label.textContent = 'Connecting to PTY…';
+
+    _ptyWs = new WebSocket(`ws://127.0.0.1:${PTY_WS_PORT}/pty`);
+    _ptyWs.binaryType = 'arraybuffer';
+
+    _ptyWs.onopen = () => {
+      if (dot) dot.className = 'ci-term-dot connected';
+      if (label) label.textContent = 'Connected — local PTY';
+      _term.writeln('\x1b[32mConnected to Nexus PTY gateway.\x1b[0m');
+      // Forward keystrokes
+      _term.onData(data => {
+        if (_ptyWs && _ptyWs.readyState === WebSocket.OPEN) {
+          _ptyWs.send(data);
+        }
+      });
+    };
+
+    _ptyWs.onmessage = (ev) => {
+      const text = (ev.data instanceof ArrayBuffer)
+        ? new TextDecoder().decode(ev.data)
+        : ev.data;
+      if (_term) _term.write(text);
+    };
+
+    _ptyWs.onclose = () => {
+      if (dot) dot.className = 'ci-term-dot disconnected';
+      if (label) label.textContent = 'Disconnected';
+      if (connectBtn) connectBtn.style.display = '';
+      if (disconnectBtn) disconnectBtn.style.display = 'none';
+      if (_term) _term.writeln('\x1b[31m[PTY connection closed]\x1b[0m');
+    };
+
+    _ptyWs.onerror = () => {
+      if (label) label.textContent = 'Connection error — is server.py running?';
+      if (dot) dot.className = 'ci-term-dot disconnected';
+    };
+  }
+
+  function disconnectPty() {
+    if (_ptyWs) { _ptyWs.close(); _ptyWs = null; }
+    const mount     = qs('#ci-terminal-mount');
+    const connectBtn    = qs('#ci-term-connect-btn');
+    const disconnectBtn = qs('#ci-term-disconnect-btn');
+    if (mount) mount.style.display = 'none';
+    if (connectBtn) connectBtn.style.display = '';
+    if (disconnectBtn) disconnectBtn.style.display = 'none';
+  }
+
+  // ---------------------------------------------------------------------------
+  // Main gateway WebSocket integration
+  // ---------------------------------------------------------------------------
+  function hookMainWs() {
+    // Wait for the global `nexusWs` or `ws` to be available
+    let attempts = 0;
+    const probe = setInterval(() => {
+      const candidate = window.nexusWs || window.ws;
+      if (candidate && candidate.readyState <= 1) {
+        _ws = candidate;
+        clearInterval(probe);
+        _ws.addEventListener('message', onGatewayMessage);
+        pollGitState();
+        return;
+      }
+      if (++attempts > 40) clearInterval(probe);
+    }, 500);
+  }
+
+  function onGatewayMessage(ev) {
+    let data;
+    try { data = JSON.parse(ev.data); } catch { return; }
+
+    switch (data.type) {
+      case 'git_status':
+        updateGitUI({
+          branch:    data.branch    || '—',
+          dirty:     !!data.dirty,
+          untracked: data.untracked || 0,
+          ahead:     data.ahead     || 0,
+        });
+        break;
+
+      case 'agent_register':
+      case 'agent_health':
+        if (data.agent_id) updateAgentHealth(data.agent_id, data);
+        break;
+
+      case 'thought':
+      case 'action':
+        if (data.agent_id) {
+          updateAgentHealth(data.agent_id, {
+            status: 'working',
+            task:   data.content || data.action || '',
+            model:  _agentHealth[data.agent_id]?.model || '',
+          });
+        }
+        break;
+
+      case 'artifact':
+        if (data.name) addArtifact({
+          name: data.name,
+          type: data.artifact_type || 'file',
+          path: data.path || '',
+          ts:   Date.now(),
+        });
+        break;
+
+      case 'memory_update':
+        if (Array.isArray(data.refs)) updateMemoryRefs(data.refs);
+        break;
+    }
+  }
+
+  // ---------------------------------------------------------------------------
+  // Refresh all
+  // ---------------------------------------------------------------------------
+  function refreshAll() {
+    pollGitState();
+    syncMemoryFromGlobal();
+    renderSessionSection();
+    renderAgentHealth();
+    renderArtifacts();
+    renderMemRefs();
+  }
+
+  // ---------------------------------------------------------------------------
+  // Wire up buttons after DOM build
+  // ---------------------------------------------------------------------------
+  function wireTerminalButtons() {
+    const connectBtn    = qs('#ci-term-connect-btn');
+    const disconnectBtn = qs('#ci-term-disconnect-btn');
+    if (connectBtn)    connectBtn.addEventListener('click', connectPty);
+    if (disconnectBtn) disconnectBtn.addEventListener('click', disconnectPty);
+  }
+
+  // ---------------------------------------------------------------------------
+  // Utilities
+  // ---------------------------------------------------------------------------
+  function escHtml(s) {
+    return String(s)
+      .replace(/&/g, '&amp;')
+      .replace(/</g, '&lt;')
+      .replace(/>/g, '&gt;')
+      .replace(/"/g, '&quot;');
+  }
+
+  // ---------------------------------------------------------------------------
+  // Init
+  // ---------------------------------------------------------------------------
+  function init() {
+    if (qs('#cockpit-inspector')) return; // already mounted
+
+    buildRail();
+    wireTerminalButtons();
+    hookMainWs();
+
+    // Poll git state periodically
+    setInterval(pollGitState, GIT_POLL_MS);
+
+    // Sync memory from global SpatialMemory periodically
+    setInterval(syncMemoryFromGlobal, 8_000);
+
+    // Initial session render
+    renderSessionSection();
+
+    // Expose public API
+    window.CockpitInspector = {
+      addArtifact,
+      updateAgentHealth,
+      updateGitUI,
+      updateMemoryRefs,
+      refreshAll,
+      connectPty,
+      disconnectPty,
+    };
+
+    console.info('[CockpitInspector] Operator rail mounted.');
+  }
+
+  // Boot after DOM is ready
+  if (document.readyState === 'loading') {
+    document.addEventListener('DOMContentLoaded', init);
+  } else {
+    // Defer slightly so app.js/boot.js finish their own init first
+    setTimeout(init, 300);
+  }
+})();

docs/ADR-001-shell-boundary.md

@@ -0,0 +1,84 @@
+# ADR-001 — Shell / Terminal Boundary and Transport Model
+
+**Status:** Accepted
+**Date:** 2026-04-22
+**Issue:** [#1695 — ATLAS Cockpit: operator inspector rail and session shell patterns](https://forge.alexanderwhitestone.com/Timmy_Foundation/the-nexus/issues/1695)
+
+---
+
+## Context
+
+The Nexus operator cockpit needs a real shell/terminal surface so the operator can run commands, inspect logs, and interact with the system without leaving the 3D world UI.
+
+Three transport models were evaluated:
+
+| Option | Description |
+|---|---|
+| **A. Native local PTY** | `server.py` spawns a PTY via Python's `pty` stdlib module; the browser connects via WebSocket on `ws://127.0.0.1:8766/pty` |
+| **B. Remote SSH PTY** | Browser connects to an SSH relay that opens a PTY on a remote machine |
+| **C. Browser pseudo-terminal** | Pure JavaScript terminal emulation (e.g., a custom REPL) with no real OS shell |
+
+---
+
+## Decision
+
+**Option A — Native local PTY** is chosen.
+
+The Nexus is explicitly a **local-first** system (CLAUDE.md: "local-first training ground for Timmy"). The operator is the same person sitting at the machine. A local PTY bridged through `server.py` is:
+
+- Architecturally consistent with the existing `server.py` WebSocket gateway
+- Zero additional infrastructure (no SSH relay, no remote server)
+- Full shell fidelity — any tool on the operator's PATH, full interactive programs, readline, color, etc.
+- Bounded: PTY_PORT (8766) binds to `127.0.0.1` only; no external exposure
+
+---
+
+## Transport Detail
+
+```
+Browser (xterm.js) ←→ ws://127.0.0.1:8766/pty ←→ server.py pty_handler ←→ OS PTY ($SHELL)
+```
+
+- Each WebSocket connection gets its own `pty.openpty()` pair and subprocess.
+- Input from xterm.js `onData` → `_ptyWs.send(data)` → `os.write(master_fd, data)`
+- Output from PTY → `os.read(master_fd)` → `websocket.send(text)` → `_term.write(text)`
+- Resize messages (`{"type":"resize","cols":N,"rows":N}`) can be added later via `fcntl.ioctl(TIOCSWINSZ)`.
+
+---
+
+## Why not Option B (Remote SSH PTY)?
+
+Remote SSH adds network hop complexity, credential management, and an SSH relay service. The Nexus does not currently have a remote operator use-case; Alexander operates locally. This decision can be revisited when fleet/remote-agent use-cases mature (see issues #672–#675).
+
+---
+
+## Why not Option C (Browser pseudo-terminal)?
+
+A JavaScript REPL cannot run arbitrary shell programs, manage processes, or provide the raw interactive shell experience that operators expect. It would be a toy, not a tool.
+
+---
+
+## Rejected Patterns
+
+- **tmux over WebSocket** — adds server-side state complexity without enough benefit at this scale
+- **ttyd** — external binary dependency; overkill for a local-first single-operator setup
+- **xterm.js + websocketd** — external binary dependency; `server.py` already owns the WebSocket gateway
+
+---
+
+## Consequences
+
+- `server.py` now starts two WebSocket servers: `PORT` (8765, main gateway) and `PTY_PORT` (8766, shell gateway)
+- `PTY_PORT` always binds to `127.0.0.1` regardless of `NEXUS_WS_HOST`
+- `cockpit-inspector.js` loads xterm.js from CDN (offline fallback: graceful degradation — the rail renders without the terminal pane)
+- PTY sessions are ephemeral (no persistence across browser reload or server restart)
+- Future: add TIOCSWINSZ resize support; add `/pty?shell=zsh` query param selection
+
+---
+
+## Related
+
+- `cockpit-inspector.js` — implements the browser side (xterm.js + WebSocket)
+- `server.py` — implements `pty_handler()` and `_get_git_status()`
+- `docs/ATLAS-COCKPIT-PATTERNS.md` — documents source patterns adopted
+- Issues: #1695, #686, #687

docs/ATLAS-COCKPIT-PATTERNS.md

@@ -0,0 +1,125 @@
+# ATLAS Cockpit — Source Patterns: Adopted, Adapted, and Rejected
+
+**Issue:** [#1695](https://forge.alexanderwhitestone.com/Timmy_Foundation/the-nexus/issues/1695)
+**Date:** 2026-04-22
+
+---
+
+## Source Repos Audited
+
+| Repo | Role in audit |
+|---|---|
+| `dodo-reach/hermes-desktop` | Primary pattern source for inspector/right rail layout |
+| `outsourc-e/hermes-workspace` | Session taxonomy vocabulary (group/tag/pin/archive) |
+| `nesquena/hermes-webui` | Session switcher UI patterns; status badge styling |
+
+---
+
+## Patterns Adopted
+
+### 1. Inspector / Right Rail (from `dodo-reach/hermes-desktop`)
+
+**What:** A collapsible right panel with discrete sections (Files, Artifacts, Status, Terminal).
+Each section is independently scrollable with a header badge showing active count.
+
+**How we adopted it:**
+- `cockpit-inspector.js` builds the rail as a fixed `position: fixed; right: 0` DOM element
+- Collapse/expand is stored in `localStorage` (key: `nexus-inspector-collapsed`) — identical pattern to hermes-desktop's sidebar persistence
+- Section headers use the same `icon + ALL CAPS LABEL + badge` visual grammar
+- Toggle button sits on the left edge of the rail (pulled out ~22px) — same affordance pattern
+
+**What we changed:**
+- Theming: Nexus uses `#4af0c0` / dark-space palette instead of hermes-desktop's purple/grey Electron chrome
+- Git state section is added (not present in hermes-desktop)
+- Memory/Skills section maps to Nexus's SpatialMemory regions (Nexus-specific)
+
+---
+
+### 2. Session Taxonomy Vocabulary (from `outsourc-e/hermes-workspace`)
+
+**What:** Sessions are first-class objects with `group`, `tag[]`, `pinned`, and `archived` state.
+Pinned sessions always sort first. Archived sessions are hidden from default lists.
+
+**How we adopted it:**
+- `session-manager.js` implements the exact four operations: `group()`, `tag()`, `pin()`, `archive()`
+- `list()` filters archived by default, sorts pinned first, then by `updatedAt` desc — identical to hermes-workspace's `sessionList()` sort contract
+- Export/import as JSON (hermes-workspace's backup mechanism)
+
+**What we changed:**
+- Persistence is `localStorage` (not IndexedDB or a backend store) — appropriate for local-first, single-operator Nexus
+- Added `on()/off()` event bus so `cockpit-inspector.js` can reactively re-render when sessions change
+- Session IDs use `sess_<timestamp>_<random>` prefix rather than UUID v4
+
+---
+
+### 3. Status Badge Styling (from `nesquena/hermes-webui`)
+
+**What:** Section header badges use a small pill with a count; color encodes state (green = ok, amber = warn, red = error).
+
+**How we adopted it:**
+- `.ci-section-badge`, `.badge-warn`, `.badge-ok` classes in `style.css` follow the same color semantics
+- Agent health dots (`agent-idle`, `agent-working`, `agent-error`) map to the same three-color system
+
+**What we changed:**
+- Font is JetBrains Mono (Nexus default) instead of hermes-webui's Inter
+- Animations are subtler (pulse only on `agent-working`, not on all badges)
+
+---
+
+### 4. xterm.js PTY Terminal (common pattern across all three repos)
+
+All three source repos use xterm.js as the browser terminal component. The transport varies:
+- hermes-desktop: IPC bridge to a native Node.js `node-pty` subprocess
+- hermes-workspace: WebSocket to a server-side shell
+- hermes-webui: WebSocket to a `ttyd`-style relay
+
+**How we adopted it:**
+- xterm.js 5.3.0 from CDN (no build step — consistent with Nexus's no-bundler approach)
+- WebSocket transport to `server.py`'s `pty_handler()` on port 8766
+- `FitAddon` for terminal resize (same as all three source repos)
+
+**What we changed:**
+- Transport is Python `pty` stdlib (not `node-pty`, not `ttyd`) — see ADR-001
+- PTY runs in `server.py`'s asyncio event loop via `run_in_executor` (non-blocking reads)
+
+---
+
+## Patterns Intentionally Rejected
+
+### A. Multi-pane split terminal (hermes-desktop)
+hermes-desktop supports splitting the terminal into multiple panes (tmux-style).
+**Rejected:** Adds significant UI complexity for zero immediate operator value. The Nexus operator uses native terminal splits in their OS. One PTY pane is sufficient.
+
+### B. Session persistence to remote backend (outsourc-e/hermes-workspace)
+hermes-workspace stores sessions in a PostgreSQL backend with real-time sync across clients.
+**Rejected:** The Nexus is local-first and single-operator. `localStorage` is sufficient and adds zero infrastructure.
+
+### C. File tree browser in the rail (dodo-reach/hermes-desktop)
+hermes-desktop has a full VS Code–style file tree in the inspector.
+**Rejected:** The Nexus 3D world is not a code editor. Artifacts are surfaced as a flat list of recently-touched files/outputs, not a tree. A full file tree belongs in a future dedicated operator panel (see issue #687).
+
+### D. Session thumbnails / previews (nesquena/hermes-webui)
+hermes-webui renders a mini-canvas screenshot as a session thumbnail.
+**Rejected:** The Nexus Three.js canvas is expensive to snapshot. Thumbnails would require canvas-capture overhead. Deferred to a future issue.
+
+### E. Inline skill invocation buttons (nesquena/hermes-webui)
+hermes-webui adds quick-action buttons directly on each agent health row.
+**Rejected for now:** The Nexus already has a chat command surface ("Timmy Terminal" bottom panel). Duplicating invocation paths would fragment the operator model. The inspector rail is read-focused; actions flow through chat.
+
+---
+
+## Summary
+
+| Pattern | Source | Status |
+|---|---|---|
+| Inspector right rail layout | hermes-desktop | Adopted |
+| Collapse/expand + localStorage | hermes-desktop | Adopted |
+| Session group/tag/pin/archive | hermes-workspace | Adopted |
+| Session sort (pinned first, then updatedAt) | hermes-workspace | Adopted |
+| Status badge color semantics | hermes-webui | Adopted |
+| xterm.js terminal | all three | Adopted (transport adapted) |
+| Multi-pane terminal | hermes-desktop | Rejected |
+| Remote session backend | hermes-workspace | Rejected |
+| File tree browser | hermes-desktop | Rejected |
+| Session thumbnails | hermes-webui | Rejected |
+| Inline skill invocation | hermes-webui | Rejected |

docs/spatial-chat-audio.md

@@ -0,0 +1,68 @@
+# Spatial Chat Audio — 3D Audio for Chat Messages
+
+Refs: the-nexus #1544
+
+## Overview
+
+Adds spatial awareness to chat notifications so nearby users/agents sound louder.
+Volume scales with avatar distance from the camera.
+
+## Features
+
+### Chat Notification Sounds
+- Each agent has a distinct tone (frequency + waveform)
+- Volume decreases with distance (inverse rolloff)
+- Stereo panning based on relative position to camera
+- Sounds auto-cleanup after playback
+
+### 3D Positional Voice (WebRTC-ready)
+- `createVoiceSource()` returns a PannerNode for real voice streams
+- HRTF panning model for realistic 3D positioning
+- Update position in real-time as avatars move
+
+### Configurable Parameters
+- `maxHearingDistance` — max distance to hear sounds (default: 40)
+- `refDistance` — full volume within this range (default: 5)
+- `rolloffFactor` — volume falloff curve (default: 1.5)
+- `baseVolume` — master volume cap (default: 0.3)
+
+## Usage
+
+```javascript
+import { SpatialChatAudio } from './nexus/components/spatial-chat-audio.js';
+
+// Initialize with camera
+SpatialChatAudio.init(camera);
+
+// Set max hearing distance
+SpatialChatAudio.setMaxHearingDistance(50);
+
+// Play a chat sound when a message arrives
+// position = avatar/agent position in 3D world
+SpatialChatAudio.playChatSound('timmy', agentPosition);
+
+// For voice chat: create a persistent 3D source
+const voice = SpatialChatAudio.createVoiceSource('user', avatarPosition);
+// Update as avatar moves
+voice.updatePosition(newPosition);
+// Cleanup when disconnected
+voice.destroy();
+```
+
+## Agent Sound Profiles
+
+| Agent  | Frequency | Waveform   |
+|--------|-----------|------------|
+| timmy  | 440 Hz    | sine       |
+| user   | 523 Hz    | sine       |
+| system | 330 Hz    | triangle   |
+| kimi   | 659 Hz    | sine       |
+| claude | 392 Hz    | sine       |
+| grok   | 587 Hz    | triangle   |
+| gemini | 494 Hz    | sine       |
+
+## Testing
+
+```bash
+node tests/test_spatial_chat_audio.js
+```

index.html

@@ -394,7 +394,16 @@
 <div id="memory-inspect-panel" class="memory-inspect-panel" style="display:none;" aria-label="Memory Inspect Panel"></div>
 <div id="memory-connections-panel" class="memory-connections-panel" style="display:none;" aria-label="Memory Connections Panel"></div>
 
+<!-- xterm.js for operator PTY terminal (cockpit inspector rail) — issue #1695 -->
+<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/xterm@5.3.0/css/xterm.css" crossorigin="anonymous">
+<script src="https://cdn.jsdelivr.net/npm/xterm@5.3.0/lib/xterm.js" crossorigin="anonymous"></script>
+<script src="https://cdn.jsdelivr.net/npm/xterm-addon-fit@0.8.0/lib/xterm-addon-fit.js" crossorigin="anonymous"></script>
+<script src="./session-manager.js"></script>
 <script src="./boot.js"></script>
+<script src="./avatar-customization.js"></script>
+<script src="./lod-system.js"></script>
+<script src="./portal-hot-reload.js"></script>
+<script src="./cockpit-inspector.js"></script>
 <script>
 function openMemoryFilter() { renderFilterList(); document.getElementById('memory-filter').style.display = 'flex'; }
 function closeMemoryFilter() { document.getElementById('memory-filter').style.display = 'none'; }

lod-system.js

@@ -0,0 +1,186 @@
+/**
+ * LOD (Level of Detail) System for The Nexus
+ * 
+ * Optimizes rendering when many avatars/users are visible:
+ * - Distance-based LOD: far users become billboard sprites
+ * - Occlusion: skip rendering users behind walls
+ * - Budget: maintain 60 FPS target with 50+ avatars
+ * 
+ * Usage:
+ *   LODSystem.init(scene, camera);
+ *   LODSystem.registerAvatar(avatarMesh, userId);
+ *   LODSystem.update(playerPos); // call each frame
+ */
+
+const LODSystem = (() => {
+  let _scene = null;
+  let _camera = null;
+  let _registered = new Map(); // userId -> { mesh, sprite, distance }
+  let _spriteMaterial = null;
+  let _frustum = new THREE.Frustum();
+  let _projScreenMatrix = new THREE.Matrix4();
+
+  // Thresholds
+  const LOD_NEAR = 15;      // Full mesh within 15 units
+  const LOD_FAR = 40;       // Billboard beyond 40 units
+  const LOD_CULL = 80;      // Don't render beyond 80 units
+  const SPRITE_SIZE = 1.2;
+
+  function init(sceneRef, cameraRef) {
+    _scene = sceneRef;
+    _camera = cameraRef;
+
+    // Create shared sprite material
+    const canvas = document.createElement('canvas');
+    canvas.width = 64;
+    canvas.height = 64;
+    const ctx = canvas.getContext('2d');
+    // Simple avatar indicator: colored circle
+    ctx.fillStyle = '#00ffcc';
+    ctx.beginPath();
+    ctx.arc(32, 32, 20, 0, Math.PI * 2);
+    ctx.fill();
+    ctx.fillStyle = '#0a0f1a';
+    ctx.beginPath();
+    ctx.arc(32, 28, 8, 0, Math.PI * 2); // head
+    ctx.fill();
+
+    const texture = new THREE.CanvasTexture(canvas);
+    _spriteMaterial = new THREE.SpriteMaterial({
+      map: texture,
+      transparent: true,
+      depthTest: true,
+      sizeAttenuation: true,
+    });
+
+    console.log('[LODSystem] Initialized');
+  }
+
+  function registerAvatar(avatarMesh, userId, color) {
+    // Create billboard sprite for this avatar
+    const spriteMat = _spriteMaterial.clone();
+    if (color) {
+      // Tint sprite to match avatar color
+      const canvas = document.createElement('canvas');
+      canvas.width = 64;
+      canvas.height = 64;
+      const ctx = canvas.getContext('2d');
+      ctx.fillStyle = color;
+      ctx.beginPath();
+      ctx.arc(32, 32, 20, 0, Math.PI * 2);
+      ctx.fill();
+      ctx.fillStyle = '#0a0f1a';
+      ctx.beginPath();
+      ctx.arc(32, 28, 8, 0, Math.PI * 2);
+      ctx.fill();
+      spriteMat.map = new THREE.CanvasTexture(canvas);
+      spriteMat.map.needsUpdate = true;
+    }
+
+    const sprite = new THREE.Sprite(spriteMat);
+    sprite.scale.set(SPRITE_SIZE, SPRITE_SIZE, 1);
+    sprite.visible = false;
+    _scene.add(sprite);
+
+    _registered.set(userId, {
+      mesh: avatarMesh,
+      sprite: sprite,
+      distance: Infinity,
+    });
+  }
+
+  function unregisterAvatar(userId) {
+    const entry = _registered.get(userId);
+    if (entry) {
+      _scene.remove(entry.sprite);
+      entry.sprite.material.dispose();
+      _registered.delete(userId);
+    }
+  }
+
+  function setSpriteColor(userId, color) {
+    const entry = _registered.get(userId);
+    if (!entry) return;
+    const canvas = document.createElement('canvas');
+    canvas.width = 64;
+    canvas.height = 64;
+    const ctx = canvas.getContext('2d');
+    ctx.fillStyle = color;
+    ctx.beginPath();
+    ctx.arc(32, 32, 20, 0, Math.PI * 2);
+    ctx.fill();
+    ctx.fillStyle = '#0a0f1a';
+    ctx.beginPath();
+    ctx.arc(32, 28, 8, 0, Math.PI * 2);
+    ctx.fill();
+    entry.sprite.material.map = new THREE.CanvasTexture(canvas);
+    entry.sprite.material.map.needsUpdate = true;
+  }
+
+  function update(playerPos) {
+    if (!_camera) return;
+
+    // Update frustum for culling
+    _projScreenMatrix.multiplyMatrices(
+      _camera.projectionMatrix,
+      _camera.matrixWorldInverse
+    );
+    _frustum.setFromProjectionMatrix(_projScreenMatrix);
+
+    _registered.forEach((entry, userId) => {
+      if (!entry.mesh) return;
+
+      const meshPos = entry.mesh.position;
+      const distance = playerPos.distanceTo(meshPos);
+      entry.distance = distance;
+
+      // Beyond cull distance: hide everything
+      if (distance > LOD_CULL) {
+        entry.mesh.visible = false;
+        entry.sprite.visible = false;
+        return;
+      }
+
+      // Check if in camera frustum
+      const inFrustum = _frustum.containsPoint(meshPos);
+      if (!inFrustum) {
+        entry.mesh.visible = false;
+        entry.sprite.visible = false;
+        return;
+      }
+
+      // LOD switching
+      if (distance <= LOD_NEAR) {
+        // Near: full mesh
+        entry.mesh.visible = true;
+        entry.sprite.visible = false;
+      } else if (distance <= LOD_FAR) {
+        // Mid: mesh with reduced detail (keep mesh visible)
+        entry.mesh.visible = true;
+        entry.sprite.visible = false;
+      } else {
+        // Far: billboard sprite
+        entry.mesh.visible = false;
+        entry.sprite.visible = true;
+        entry.sprite.position.copy(meshPos);
+        entry.sprite.position.y += 1.2; // above avatar center
+      }
+    });
+  }
+
+  function getStats() {
+    let meshCount = 0;
+    let spriteCount = 0;
+    let culledCount = 0;
+    _registered.forEach(entry => {
+      if (entry.mesh.visible) meshCount++;
+      else if (entry.sprite.visible) spriteCount++;
+      else culledCount++;
+    });
+    return { total: _registered.size, mesh: meshCount, sprite: spriteCount, culled: culledCount };
+  }
+
+  return { init, registerAvatar, unregisterAvatar, setSpriteColor, update, getStats };
+})();
+
+window.LODSystem = LODSystem;

mempalace/rooms.yaml

@@ -62,6 +62,15 @@ core_rooms:
       - proof-of-concept code snippets
       - benchmark data
 
+  - key: sovereign
+    label: Sovereign
+    purpose: Artifacts of Alexander Whitestone's requests, directives, and wizard responses
+    examples:
+      - dated request/response artifacts
+      - conversation summaries with speaker tags
+      - directive ledgers
+      - response follow-through notes
+
 optional_rooms:
   - key: evennia
     label: Evennia
@@ -98,15 +107,6 @@ optional_rooms:
     purpose: Catch-all for artefacts not yet assigned to a named room
     wizards: ["*"]
 
-  - key: sovereign
-    label: Sovereign
-    purpose: Artifacts of Alexander Whitestone's requests, directives, and conversation history
-    wizards: ["*"]
-    conventions:
-      naming: "YYYY-MM-DD_HHMMSS_<topic>.md"
-      index: "INDEX.md"
-      description: "Each artifact is a dated record of a request from Alexander and the wizard's response. The running INDEX.md provides a chronological catalog."
-
 # Tunnel routing table
 # Defines which room pairs are connected across wizard wings.
 # A tunnel lets `recall <query> --fleet` search both wings at once.

nexus/__init__.py

@@ -14,6 +14,7 @@ from nexus.perception_adapter import (
 )
 from nexus.experience_store import ExperienceStore
 from nexus.trajectory_logger import TrajectoryLogger
+from nexus.chronicle import ChronicleWriter, AgentEvent, EventKind
 
 try:
     from nexus.nexus_think import NexusMind
@@ -29,4 +30,7 @@ __all__ = [
     "ExperienceStore",
     "TrajectoryLogger",
     "NexusMind",
+    "ChronicleWriter",
+    "AgentEvent",
+    "EventKind",
 ]

nexus/bannerlord_harness.py

@@ -29,7 +29,7 @@ from typing import Any, Callable, Optional
 
 import websockets
 
-from bannerlord_trace import BannerlordTraceLogger
+from nexus.bannerlord_trace import BannerlordTraceLogger
 
 # ═══════════════════════════════════════════════════════════════════════════
 # CONFIGURATION

nexus/chronicle.py

@@ -0,0 +1,387 @@
+"""
+Nexus Chronicle — Emergent Narrative from Agent Interactions
+
+Watches the fleet's activity (dispatches, errors, recoveries,
+collaborations) and transforms raw event data into narrative prose.
+The system finds the dramatic arc in real work and produces a living
+chronicle. The story writes itself from the data.
+
+Usage:
+    from nexus.chronicle import ChronicleWriter, AgentEvent, EventKind
+
+    writer = ChronicleWriter()
+
+    writer.ingest(AgentEvent(kind=EventKind.DISPATCH, agent="claude", detail="took issue #42"))
+    writer.ingest(AgentEvent(kind=EventKind.ERROR, agent="claude", detail="rate limit hit"))
+    writer.ingest(AgentEvent(kind=EventKind.RECOVERY, agent="claude", detail="retried after backoff"))
+    writer.ingest(AgentEvent(kind=EventKind.COMMIT, agent="claude", detail="feat: add narrative engine"))
+
+    prose = writer.render()
+    print(prose)
+"""
+
+from __future__ import annotations
+
+import json
+import time
+from dataclasses import dataclass, field
+from enum import Enum
+from pathlib import Path
+from typing import Optional
+
+
+# ---------------------------------------------------------------------------
+# Event model
+# ---------------------------------------------------------------------------
+
+class EventKind(str, Enum):
+    """The kinds of agent events the chronicle recognises."""
+
+    DISPATCH = "dispatch"         # agent claimed / was assigned work
+    COMMIT = "commit"             # agent produced a commit
+    PUSH = "push"                 # agent pushed a branch
+    PR_OPEN = "pr_open"           # agent opened a pull request
+    PR_MERGE = "pr_merge"         # PR was merged
+    ERROR = "error"               # agent hit an error / exception
+    RECOVERY = "recovery"         # agent recovered from a failure
+    ABANDON = "abandon"           # agent abandoned a task (timeout / giving up)
+    COLLABORATION = "collab"      # two agents worked on the same thing
+    HEARTBEAT = "heartbeat"       # agent reported a heartbeat (alive signal)
+    IDLE = "idle"                 # agent is waiting for work
+    MILESTONE = "milestone"       # notable achievement (e.g. 100th issue closed)
+
+
+@dataclass
+class AgentEvent:
+    """One discrete thing that happened in the fleet."""
+
+    kind: EventKind
+    agent: str                          # who did this (e.g. "claude", "mimo-v2-pro")
+    detail: str = ""                    # free-text description
+    timestamp: float = field(default_factory=time.time)
+    metadata: dict = field(default_factory=dict)
+
+    def to_dict(self) -> dict:
+        return {
+            "kind": self.kind.value,
+            "agent": self.agent,
+            "detail": self.detail,
+            "timestamp": self.timestamp,
+            "metadata": self.metadata,
+        }
+
+    @classmethod
+    def from_dict(cls, data: dict) -> "AgentEvent":
+        return cls(
+            kind=EventKind(data["kind"]),
+            agent=data["agent"],
+            detail=data.get("detail", ""),
+            timestamp=data.get("timestamp", time.time()),
+            metadata=data.get("metadata", {}),
+        )
+
+
+# ---------------------------------------------------------------------------
+# Narrative templates — maps event kinds to prose fragments
+# ---------------------------------------------------------------------------
+
+# Each entry is a list so we can rotate through variants.
+_TEMPLATES: dict[EventKind, list[str]] = {
+    EventKind.DISPATCH: [
+        "{agent} stepped forward and claimed the work: {detail}.",
+        "{agent} took on the challenge — {detail}.",
+        "The task landed on {agent}'s desk: {detail}.",
+    ],
+    EventKind.COMMIT: [
+        '{agent} sealed a commit into the record: "{detail}".',
+        '{agent} committed "{detail}" — progress crystallised.',
+        "{agent} carved a new ring into the trunk: {detail}.",
+    ],
+    EventKind.PUSH: [
+        "{agent} pushed the work upstream.",
+        "The branch rose into the forge — {agent}'s changes were live.",
+        "{agent} sent their work into the wider current.",
+    ],
+    EventKind.PR_OPEN: [
+        "{agent} opened a pull request: {detail}.",
+        "A proposal surfaced — {agent} asked the fleet to review {detail}.",
+        "{agent} laid their work before the reviewers: {detail}.",
+    ],
+    EventKind.PR_MERGE: [
+        "{agent}'s branch folded into the whole: {detail}.",
+        "Consensus reached — {agent}'s changes were merged: {detail}.",
+        "{detail} joined the canon. {agent}'s contribution lives on.",
+    ],
+    EventKind.ERROR: [
+        "{agent} ran into an obstacle: {detail}.",
+        "Trouble. {agent} encountered {detail} and had to pause.",
+        "The path grew difficult — {agent} hit {detail}.",
+    ],
+    EventKind.RECOVERY: [
+        "{agent} regrouped and pressed on: {detail}.",
+        "After the setback, {agent} found a way through: {detail}.",
+        "{agent} recovered — {detail}.",
+    ],
+    EventKind.ABANDON: [
+        "{agent} released the task, unable to finish: {detail}.",
+        "Sometimes wisdom is knowing when to let go. {agent} abandoned {detail}.",
+        "{agent} stepped back from {detail}. Another will carry it forward.",
+    ],
+    EventKind.COLLABORATION: [
+        "{agent} and their peers converged on the same problem: {detail}.",
+        "Two minds touched the same work — {agent} in collaboration: {detail}.",
+        "The fleet coordinated — {agent} joined the effort on {detail}.",
+    ],
+    EventKind.HEARTBEAT: [
+        "{agent} checked in — still thinking, still present.",
+        "A pulse from {agent}: the mind is alive.",
+        "{agent} breathed through another cycle.",
+    ],
+    EventKind.IDLE: [
+        "{agent} rested, waiting for the next call.",
+        "Quiet descended — {agent} held still between tasks.",
+        "{agent} stood ready, watchful in the lull.",
+    ],
+    EventKind.MILESTONE: [
+        "A moment worth noting — {agent}: {detail}.",
+        "The chronicle marks a milestone. {agent}: {detail}.",
+        "History ticked over — {agent} reached {detail}.",
+    ],
+}
+
+# Arc-level commentary triggered by sequences of events
+_ARC_TEMPLATES = {
+    "struggle_and_recovery": (
+        "There was a struggle here. {agent} hit trouble and came back stronger — "
+        "the kind of arc that gives a chronicle its texture."
+    ),
+    "silent_grind": (
+        "No drama, just steady work. {agents} moved through the backlog with quiet persistence."
+    ),
+    "abandon_then_retry": (
+        "{agent} let go once. But the work called again, and this time it was answered."
+    ),
+    "solo_sprint": (
+        "{agent} ran the whole arc alone — dispatch to merge — without breaking stride."
+    ),
+    "fleet_convergence": (
+        "The fleet converged. Multiple agents touched the same thread and wove it tighter."
+    ),
+}
+
+
+# ---------------------------------------------------------------------------
+# Chronicle writer
+# ---------------------------------------------------------------------------
+
+class ChronicleWriter:
+    """Accumulates agent events and renders them as narrative prose.
+
+    The writer keeps a running log of events.  Call ``ingest()`` to add new
+    events as they arrive, then ``render()`` to produce a prose snapshot of
+    the current arc.
+
+    Events are also persisted to JSONL so the chronicle survives restarts.
+    """
+
+    def __init__(self, log_path: Optional[Path] = None):
+        today = time.strftime("%Y-%m-%d")
+        self.log_path = log_path or (
+            Path.home() / ".nexus" / "chronicle" / f"chronicle_{today}.jsonl"
+        )
+        self.log_path.parent.mkdir(parents=True, exist_ok=True)
+
+        self._events: list[AgentEvent] = []
+        self._template_counters: dict[EventKind, int] = {}
+
+        # Load any events already on disk for today
+        self._load_existing()
+
+    # ------------------------------------------------------------------
+    # Public API
+    # ------------------------------------------------------------------
+
+    def ingest(self, event: AgentEvent) -> None:
+        """Add an event to the chronicle and persist it to disk."""
+        self._events.append(event)
+        with open(self.log_path, "a") as f:
+            f.write(json.dumps(event.to_dict()) + "\n")
+
+    def render(self, max_events: int = 50) -> str:
+        """Render the recent event stream as narrative prose.
+
+        Returns a multi-paragraph string suitable for display or logging.
+        """
+        events = self._events[-max_events:]
+        if not events:
+            return "The chronicle is empty. No events have been recorded yet."
+
+        paragraphs: list[str] = []
+
+        # Opening line with timestamp range
+        first_ts = time.strftime("%H:%M", time.localtime(events[0].timestamp))
+        last_ts = time.strftime("%H:%M", time.localtime(events[-1].timestamp))
+        paragraphs.append(
+            f"The chronicle covers {len(events)} event(s) between {first_ts} and {last_ts}."
+        )
+
+        # Event-by-event prose
+        sentences: list[str] = []
+        for evt in events:
+            sentences.append(self._render_event(evt))
+        paragraphs.append("  ".join(sentences))
+
+        # Arc-level commentary
+        arc = self._detect_arc(events)
+        if arc:
+            paragraphs.append(arc)
+
+        return "\n\n".join(paragraphs)
+
+    def render_markdown(self, max_events: int = 50) -> str:
+        """Render as a Markdown document."""
+        events = self._events[-max_events:]
+        if not events:
+            return "# Chronicle\n\n*No events recorded yet.*"
+
+        today = time.strftime("%Y-%m-%d")
+        lines = [f"# Chronicle — {today}", ""]
+
+        for evt in events:
+            ts = time.strftime("%H:%M:%S", time.localtime(evt.timestamp))
+            prose = self._render_event(evt)
+            lines.append(f"**{ts}** — {prose}")
+
+        arc = self._detect_arc(events)
+        if arc:
+            lines += ["", "---", "", f"*{arc}*"]
+
+        return "\n".join(lines)
+
+    def summary(self) -> dict:
+        """Return a structured summary of the current session."""
+        agents: dict[str, dict] = {}
+        kind_counts: dict[str, int] = {}
+
+        for evt in self._events:
+            agents.setdefault(evt.agent, {"events": 0, "kinds": []})
+            agents[evt.agent]["events"] += 1
+            agents[evt.agent]["kinds"].append(evt.kind.value)
+            kind_counts[evt.kind.value] = kind_counts.get(evt.kind.value, 0) + 1
+
+        return {
+            "total_events": len(self._events),
+            "agents": agents,
+            "kind_counts": kind_counts,
+            "log_path": str(self.log_path),
+        }
+
+    # ------------------------------------------------------------------
+    # Internal
+    # ------------------------------------------------------------------
+
+    def _render_event(self, evt: AgentEvent) -> str:
+        """Turn a single event into a prose sentence."""
+        templates = _TEMPLATES.get(evt.kind, ["{agent}: {detail}"])
+        counter = self._template_counters.get(evt.kind, 0)
+        template = templates[counter % len(templates)]
+        self._template_counters[evt.kind] = counter + 1
+        return template.format(agent=evt.agent, detail=evt.detail or evt.kind.value)
+
+    def _detect_arc(self, events: list[AgentEvent]) -> Optional[str]:
+        """Scan the event sequence for a recognisable dramatic arc."""
+        if not events:
+            return None
+
+        kinds = [e.kind for e in events]
+        agents = list({e.agent for e in events})
+
+        # struggle → recovery
+        if EventKind.ERROR in kinds and EventKind.RECOVERY in kinds:
+            err_idx = kinds.index(EventKind.ERROR)
+            rec_idx = kinds.index(EventKind.RECOVERY)
+            if rec_idx > err_idx:
+                agent = events[err_idx].agent
+                return _ARC_TEMPLATES["struggle_and_recovery"].format(agent=agent)
+
+        # abandon → dispatch (retry): find first ABANDON, then any DISPATCH after it
+        if EventKind.ABANDON in kinds and EventKind.DISPATCH in kinds:
+            ab_idx = kinds.index(EventKind.ABANDON)
+            retry_idx = next(
+                (i for i, k in enumerate(kinds) if k == EventKind.DISPATCH and i > ab_idx),
+                None,
+            )
+            if retry_idx is not None:
+                agent = events[retry_idx].agent
+                return _ARC_TEMPLATES["abandon_then_retry"].format(agent=agent)
+
+        # solo sprint: single agent goes dispatch→commit→pr_open→pr_merge
+        solo_arc = {EventKind.DISPATCH, EventKind.COMMIT, EventKind.PR_OPEN, EventKind.PR_MERGE}
+        if solo_arc.issubset(set(kinds)) and len(agents) == 1:
+            return _ARC_TEMPLATES["solo_sprint"].format(agent=agents[0])
+
+        # fleet convergence: multiple agents, collaboration event
+        if len(agents) > 1 and EventKind.COLLABORATION in kinds:
+            return _ARC_TEMPLATES["fleet_convergence"]
+
+        # silent grind: only commits / heartbeats, no drama
+        drama = {EventKind.ERROR, EventKind.ABANDON, EventKind.RECOVERY, EventKind.COLLABORATION}
+        if not drama.intersection(set(kinds)) and EventKind.COMMIT in kinds:
+            return _ARC_TEMPLATES["silent_grind"].format(agents=", ".join(agents))
+
+        return None
+
+    def _load_existing(self) -> None:
+        """Load events persisted from earlier in the same session."""
+        if not self.log_path.exists():
+            return
+        with open(self.log_path) as f:
+            for line in f:
+                line = line.strip()
+                if not line:
+                    continue
+                try:
+                    self._events.append(AgentEvent.from_dict(json.loads(line)))
+                except (json.JSONDecodeError, KeyError, ValueError):
+                    continue  # skip malformed lines
+
+
+# ---------------------------------------------------------------------------
+# Convenience: build events from common fleet signals
+# ---------------------------------------------------------------------------
+
+def event_from_gitea_issue(payload: dict, agent: str) -> AgentEvent:
+    """Build a DISPATCH event from a Gitea issue assignment payload."""
+    issue_num = payload.get("number", "?")
+    title = payload.get("title", "")
+    return AgentEvent(
+        kind=EventKind.DISPATCH,
+        agent=agent,
+        detail=f"issue #{issue_num}: {title}",
+        metadata={"issue_number": issue_num},
+    )
+
+
+def event_from_heartbeat(hb: dict) -> AgentEvent:
+    """Build a HEARTBEAT event from a nexus heartbeat dict."""
+    agent = hb.get("model", "unknown")
+    status = hb.get("status", "thinking")
+    cycle = hb.get("cycle", 0)
+    return AgentEvent(
+        kind=EventKind.HEARTBEAT,
+        agent=agent,
+        detail=f"cycle {cycle}, status={status}",
+        metadata=hb,
+    )
+
+
+def event_from_commit(commit: dict, agent: str) -> AgentEvent:
+    """Build a COMMIT event from a git commit dict."""
+    message = commit.get("message", "").split("\n")[0]  # subject line only
+    sha = commit.get("sha", "")[:8]
+    return AgentEvent(
+        kind=EventKind.COMMIT,
+        agent=agent,
+        detail=message,
+        metadata={"sha": sha},
+    )

nexus/components/spatial-chat-audio.js

@@ -0,0 +1,236 @@
+// ════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════
+//  SPATIAL CHAT AUDIO — 3D Audio for Chat Messages (issue #1544)
+// ════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════
+//
+// Volume scales with avatar distance — closer agents sound louder.
+// 3D positional audio places chat sounds in world space.
+//
+// Usage from app.js:
+//   import { SpatialChatAudio } from './nexus/components/spatial-chat-audio.js';
+//   SpatialChatAudio.init(camera);
+//   SpatialChatAudio.playChatSound('timmy', agentPosition);
+//
+// Configuration:
+//   SpatialChatAudio.setMaxHearingDistance(50); // default 40 units
+//   SpatialChatAudio.setEnabled(true/false);
+// ═══════════════════════════════════════════════════════════════════════════════════════════════════════════════════════════
+
+const SpatialChatAudio = (() => {
+
+  // ─── CONFIG ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
+  let _config = {
+    maxHearingDistance: 40,   // Distance at which volume reaches 0
+    refDistance: 5,           // Full volume within this range
+    rolloffFactor: 1.5,       // Volume rolloff curve
+    baseVolume: 0.3,          // Master volume for chat sounds
+    enabled: true,
+  };
+
+  // ─── STATE ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
+  let _camera = null;
+  let _listener = null;
+  let _ctx = null;
+  let _masterGain = null;
+  let _initialized = false;
+
+  // Agent sound profiles (frequency + waveform)
+  const AGENT_SOUNDS = {
+    timmy:    { freq: 440, type: 'sine' },      // A4 - clear
+    user:     { freq: 523, type: 'sine' },      // C5 - higher
+    system:   { freq: 330, type: 'triangle' },  // E4 - neutral
+    kimi:     { freq: 659, type: 'sine' },      // E5 - bright
+    claude:   { freq: 392, type: 'sine' },      // G4 - warm
+    grok:     { freq: 587, type: 'triangle' },  // D5 - sharp
+    gemini:   { freq: 494, type: 'sine' },      // B4 - balanced
+    default:  { freq: 440, type: 'sine' },      // A4 - default
+  };
+
+  // ─── INIT ──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
+  function init(camera) {
+    if (_initialized) return;
+
+    _camera = camera;
+
+    // Create or reuse AudioListener from camera
+    if (!_listener) {
+      _listener = new THREE.AudioListener();
+      camera.add(_listener);
+    } else {
+      _listener = camera.getObjectByProperty('type', 'AudioListener') || _listener;
+    }
+
+    // Get audio context from listener
+    _ctx = _listener.context;
+    _masterGain = _ctx.createGain();
+    _masterGain.gain.value = _config.baseVolume;
+    _masterGain.connect(_ctx.destination);
+
+    _initialized = true;
+    console.info('[SpatialChatAudio] Initialized — max hearing distance:', _config.maxHearingDistance);
+
+    // Resume context if suspended (browser autoplay policy)
+    if (_ctx.state === 'suspended') {
+      const resume = () => {
+        _ctx.resume().then(() => {
+          console.info('[SpatialChatAudio] AudioContext resumed');
+          document.removeEventListener('click', resume);
+          document.removeEventListener('keydown', resume);
+        });
+      };
+      document.addEventListener('click', resume);
+      document.addEventListener('keydown', resume);
+    }
+
+    return _listener;
+  }
+
+  // ─── PLAY CHAT SOUND ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
+  function playChatSound(agent, position) {
+    if (!_initialized || !_config.enabled) return;
+
+    const sound = AGENT_SOUNDS[agent] || AGENT_SOUNDS.default;
+    const camPos = _camera.position;
+
+    // Calculate distance
+    const dx = position.x - camPos.x;
+    const dy = position.y - camPos.y;
+    const dz = position.z - camPos.z;
+    const dist = Math.sqrt(dx * dx + dy * dy + dz * dz);
+
+    // Calculate volume based on distance
+    let volume = 0;
+    if (dist < _config.maxHearingDistance) {
+      volume = 1 / (1 + _config.rolloffFactor * (dist - _config.refDistance));
+      volume = Math.max(0, Math.min(1, volume));
+    }
+
+    // Skip if too quiet
+    if (volume < 0.01) return;
+
+    // Create audio nodes
+    const osc = _ctx.createOscillator();
+    osc.type = sound.type;
+    osc.frequency.value = sound.freq;
+
+    const gain = _ctx.createGain();
+    gain.gain.value = volume;
+
+    const panner = _ctx.createStereoPanner();
+
+    // Calculate stereo panning
+    const camRight = new THREE.Vector3();
+    _camera.getWorldDirection(camRight);
+    camRight.cross(_camera.up).normalize();
+    const toSource = new THREE.Vector3(dx, 0, dz).normalize();
+    const pan = THREE.MathUtils.clamp(toSource.dot(camRight), -1, 1);
+    panner.pan.value = pan;
+
+    // Connect and play
+    osc.connect(gain);
+    gain.connect(panner);
+    panner.connect(_masterGain);
+
+    // Short envelope (attack + decay)
+    const now = _ctx.currentTime;
+    gain.gain.setValueAtTime(0, now);
+    gain.gain.linearRampToValueAtTime(volume, now + 0.01);
+    gain.gain.exponentialRampToValueAtTime(0.001, now + 0.3);
+
+    osc.start(now);
+    osc.stop(now + 0.35);
+
+    // Cleanup
+    osc.onended = () => {
+      osc.disconnect();
+      gain.disconnect();
+      panner.disconnect();
+    };
+
+    console.debug(`[SpatialChatAudio] ${agent} at ${dist.toFixed(1)}m, vol=${volume.toFixed(2)}, pan=${pan.toFixed(2)}`);
+  }
+
+  // ─── CONFIGURATION ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
+  function setMaxHearingDistance(dist) {
+    _config.maxHearingDistance = Math.max(5, dist);
+    console.info('[SpatialChatAudio] Max hearing distance:', _config.maxHearingDistance);
+  }
+
+  function getMaxHearingDistance() {
+    return _config.maxHearingDistance;
+  }
+
+  function setEnabled(enabled) {
+    _config.enabled = enabled;
+    console.info('[SpatialChatAudio]', enabled ? 'Enabled' : 'Disabled');
+  }
+
+  function isEnabled() {
+    return _config.enabled;
+  }
+
+  function setMasterVolume(vol) {
+    if (_masterGain) {
+      _masterGain.gain.setTargetAtTime(
+        THREE.MathUtils.clamp(vol, 0, 1),
+        _ctx.currentTime,
+        0.05
+      );
+    }
+  }
+
+  // ─── VOICE CHAT SUPPORT (WebRTC placeholder) ─────────────────────────────────────────────────────────────────────────────────────────
+  function createVoiceSource(agentId, position) {
+    if (!_initialized) return null;
+
+    // Create a PannerNode for 3D voice positioning
+    const panner = _ctx.createPanner();
+    panner.panningModel = 'HRTF';
+    panner.distanceModel = 'inverse';
+    panner.refDistance = _config.refDistance;
+    panner.maxDistance = _config.maxHearingDistance;
+    panner.rolloffFactor = _config.rolloffFactor;
+
+    // Set initial position
+    panner.positionX.value = position.x;
+    panner.positionY.value = position.y;
+    panner.positionZ.value = position.z;
+
+    // Connect to master
+    panner.connect(_masterGain);
+
+    console.info(`[SpatialChatAudio] Voice source created for ${agentId}`);
+
+    return {
+      panner,
+      agentId,
+      updatePosition(pos) {
+        panner.positionX.setValueAtTime(pos.x, _ctx.currentTime);
+        panner.positionY.setValueAtTime(pos.y, _ctx.currentTime);
+        panner.positionZ.setValueAtTime(pos.z, _ctx.currentTime);
+      },
+      destroy() {
+        panner.disconnect();
+        console.info(`[SpatialChatAudio] Voice source destroyed for ${agentId}`);
+      }
+    };
+  }
+
+  // ─── API ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
+  return {
+    init,
+    playChatSound,
+    createVoiceSource,
+    setMaxHearingDistance,
+    getMaxHearingDistance,
+    setEnabled,
+    isEnabled,
+    setMasterVolume,
+  };
+})();
+
+// Export for module or global usage
+if (typeof module !== 'undefined' && module.exports) {
+  module.exports = { SpatialChatAudio };
+} else if (typeof window !== 'undefined') {
+  window.SpatialChatAudio = SpatialChatAudio;
+}

nexus/evennia_ws_bridge.py

@@ -304,6 +304,43 @@ async def inject_event(event_type: str, ws_url: str, **kwargs):
         sys.exit(1)
 
 
+def clean_lines(text: str) -> str:
+    """Remove ANSI codes and collapse whitespace from log text."""
+    import re
+    text = strip_ansi(text)
+    text = re.sub(r'\s+', ' ', text).strip()
+    return text
+
+
+def normalize_event(event: dict) -> dict:
+    """Normalize an Evennia event dict to standard format."""
+    return {
+        "type": event.get("type", "unknown"),
+        "actor": event.get("actor", event.get("name", "")),
+        "room": event.get("room", event.get("location", "")),
+        "message": event.get("message", event.get("text", "")),
+        "timestamp": event.get("timestamp", ""),
+    }
+
+
+def parse_room_output(text: str) -> dict:
+    """Parse Evennia room output into structured data."""
+    import re
+    lines = text.strip().split("\n")
+    result = {"name": "", "description": "", "exits": [], "objects": []}
+    if lines:
+        result["name"] = strip_ansi(lines[0]).strip()
+    if len(lines) > 1:
+        result["description"] = strip_ansi(lines[1]).strip()
+    for line in lines[2:]:
+        line = strip_ansi(line).strip()
+        if line.startswith("Exits:"):
+            result["exits"] = [e.strip() for e in line[6:].split(",") if e.strip()]
+        elif line.startswith("You see:"):
+            result["objects"] = [o.strip() for o in line[8:].split(",") if o.strip()]
+    return result
+
+
 def main():
     parser = argparse.ArgumentParser(description="Evennia -> Nexus WebSocket Bridge")
     sub = parser.add_subparsers(dest="mode")

nexus/mcdonald_wizard.py

@@ -0,0 +1,283 @@
+#!/usr/bin/env python3
+"""
+McDonald Wizard — Hermes shim for the McDonald chatbot API
+
+Exposes the `mcdonald-wizard` Hermes tool, which forwards prompts to the
+McDonald chatbot API and returns wizard-style responses. Registered as a
+Hermes skill via ~/.hermes/skills/shim-mcdonald-wizard.py.
+
+Usage:
+    from nexus.mcdonald_wizard import McdonaldWizard
+    wizard = McdonaldWizard()
+    response = wizard.ask("What is your quest?")
+    print(response.text)
+
+Environment Variables:
+    MCDONALDS_API_KEY   — McDonald chatbot API key (required)
+    MCDONALDS_ENDPOINT  — API endpoint (default: https://api.mcdonalds.com/v1/chat)
+    MCDONALDS_TIMEOUT   — Request timeout in seconds (default: 30)
+    MCDONALDS_RETRIES   — Max retry attempts (default: 3)
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+import time
+from dataclasses import dataclass, field
+from datetime import datetime, timezone
+from typing import Optional
+
+import requests
+
+log = logging.getLogger("mcdonald_wizard")
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s [mcdonald_wizard] %(message)s",
+    datefmt="%H:%M:%S",
+)
+
+DEFAULT_ENDPOINT = "https://api.mcdonalds.com/v1/chat"
+DEFAULT_TIMEOUT = 30
+DEFAULT_RETRIES = 3
+WIZARD_ID = "mcdonald-wizard"
+
+# Retry backoff: base * 2^(attempt-1)
+RETRY_BASE_DELAY = 1.0
+
+
+@dataclass
+class WizardResponse:
+    """Response from the McDonald chatbot wizard."""
+
+    text: str = ""
+    model: str = ""
+    latency_ms: float = 0.0
+    attempt: int = 1
+    error: Optional[str] = None
+    timestamp: str = field(
+        default_factory=lambda: datetime.now(timezone.utc).isoformat()
+    )
+
+    def to_dict(self) -> dict:
+        return {
+            "text": self.text,
+            "model": self.model,
+            "latency_ms": self.latency_ms,
+            "attempt": self.attempt,
+            "error": self.error,
+            "timestamp": self.timestamp,
+        }
+
+
+class McdonaldWizard:
+    """
+    McDonald chatbot wizard client.
+
+    Forwards prompts to the McDonald chatbot API with retry/timeout handling.
+    Integrates with Hermes as the `mcdonald-wizard` tool.
+    """
+
+    def __init__(
+        self,
+        api_key: Optional[str] = None,
+        endpoint: Optional[str] = None,
+        timeout: Optional[int] = None,
+        max_retries: Optional[int] = None,
+    ):
+        self.api_key = api_key or os.environ.get("MCDONALDS_API_KEY", "")
+        self.endpoint = endpoint or os.environ.get(
+            "MCDONALDS_ENDPOINT", DEFAULT_ENDPOINT
+        )
+        self.timeout = timeout or int(
+            os.environ.get("MCDONALDS_TIMEOUT", DEFAULT_TIMEOUT)
+        )
+        self.max_retries = max_retries or int(
+            os.environ.get("MCDONALDS_RETRIES", DEFAULT_RETRIES)
+        )
+
+        if not self.api_key:
+            log.warning(
+                "MCDONALDS_API_KEY not set — wizard will return errors on live calls"
+            )
+
+        # Session stats
+        self.request_count = 0
+        self.total_latency_ms = 0.0
+
+    def _headers(self) -> dict:
+        return {
+            "Authorization": f"Bearer {self.api_key}",
+            "Content-Type": "application/json",
+        }
+
+    def _post_with_retry(self, payload: dict) -> tuple[dict, int, float]:
+        """
+        POST to the McDonald API with retry/backoff.
+
+        Returns (response_json, attempt_number, latency_ms).
+        Raises on final failure.
+        """
+        last_exc: Optional[Exception] = None
+        for attempt in range(1, self.max_retries + 1):
+            t0 = time.monotonic()
+            try:
+                resp = requests.post(
+                    self.endpoint,
+                    json=payload,
+                    headers=self._headers(),
+                    timeout=self.timeout,
+                )
+                latency_ms = (time.monotonic() - t0) * 1000
+                if resp.status_code in (429, 500, 502, 503, 504):
+                    raise requests.HTTPError(
+                        f"HTTP {resp.status_code}: {resp.text[:200]}"
+                    )
+                resp.raise_for_status()
+                return resp.json(), attempt, latency_ms
+            except Exception as exc:
+                last_exc = exc
+                if attempt < self.max_retries:
+                    delay = RETRY_BASE_DELAY * (2 ** (attempt - 1))
+                    log.warning(
+                        "attempt %d/%d failed (%s) — retrying in %.1fs",
+                        attempt,
+                        self.max_retries,
+                        exc,
+                        delay,
+                    )
+                    time.sleep(delay)
+                else:
+                    log.error(
+                        "all %d attempts failed: %s", self.max_retries, exc
+                    )
+        raise last_exc  # type: ignore[misc]
+
+    def ask(
+        self,
+        prompt: str,
+        system: Optional[str] = None,
+        context: Optional[str] = None,
+    ) -> WizardResponse:
+        """
+        Send a prompt to the McDonald wizard chatbot.
+
+        Args:
+            prompt:  User message to the wizard.
+            system:  Optional system instruction override.
+            context: Optional prior context to prepend.
+
+        Returns:
+            WizardResponse with text, latency, and error fields.
+        """
+        if not self.api_key:
+            return WizardResponse(
+                error="MCDONALDS_API_KEY not set — cannot call McDonald wizard API"
+            )
+
+        messages = []
+        if system:
+            messages.append({"role": "system", "content": system})
+        if context:
+            messages.append({"role": "user", "content": context})
+            messages.append(
+                {"role": "assistant", "content": "Understood, I have the context."}
+            )
+        messages.append({"role": "user", "content": prompt})
+
+        payload = {"messages": messages}
+
+        t0 = time.monotonic()
+        try:
+            data, attempt, latency_ms = self._post_with_retry(payload)
+        except Exception as exc:
+            latency_ms = (time.monotonic() - t0) * 1000
+            self.request_count += 1
+            self.total_latency_ms += latency_ms
+            return WizardResponse(
+                error=f"McDonald wizard API failed: {exc}",
+                latency_ms=latency_ms,
+            )
+
+        self.request_count += 1
+        self.total_latency_ms += latency_ms
+
+        text = (
+            data.get("choices", [{}])[0]
+            .get("message", {})
+            .get("content", "")
+        )
+        model = data.get("model", "")
+
+        return WizardResponse(
+            text=text,
+            model=model,
+            latency_ms=latency_ms,
+            attempt=attempt,
+        )
+
+    def session_stats(self) -> dict:
+        """Return session telemetry."""
+        return {
+            "wizard_id": WIZARD_ID,
+            "request_count": self.request_count,
+            "total_latency_ms": self.total_latency_ms,
+            "avg_latency_ms": (
+                self.total_latency_ms / self.request_count
+                if self.request_count
+                else 0.0
+            ),
+        }
+
+
+# ── Hermes tool function ──────────────────────────────────────────────────
+
+_wizard_instance: Optional[McdonaldWizard] = None
+
+
+def _get_wizard() -> McdonaldWizard:
+    global _wizard_instance
+    if _wizard_instance is None:
+        _wizard_instance = McdonaldWizard()
+    return _wizard_instance
+
+
+def mcdonald_wizard(prompt: str, system: Optional[str] = None) -> dict:
+    """
+    Hermes tool: forward *prompt* to the McDonald chatbot wizard.
+
+    Args:
+        prompt: The message to send to the wizard.
+        system: Optional system instruction.
+
+    Returns:
+        dict with keys: text, model, latency_ms, attempt, error.
+    """
+    wizard = _get_wizard()
+    resp = wizard.ask(prompt, system=system)
+    return resp.to_dict()
+
+
+# ── CLI ───────────────────────────────────────────────────────────────────
+
+
+def main() -> None:
+    import argparse
+
+    parser = argparse.ArgumentParser(description="McDonald Wizard CLI")
+    parser.add_argument("prompt", nargs="?", default="Greetings, wizard!", help="Prompt to send")
+    parser.add_argument("--system", default=None, help="System instruction")
+    parser.add_argument("--endpoint", default=None, help="API endpoint override")
+    args = parser.parse_args()
+
+    wizard = McdonaldWizard(endpoint=args.endpoint)
+    resp = wizard.ask(args.prompt, system=args.system)
+    if resp.error:
+        print(f"[ERROR] {resp.error}")
+    else:
+        print(resp.text)
+        print(f"\n[latency={resp.latency_ms:.0f}ms attempt={resp.attempt} model={resp.model}]")
+
+
+if __name__ == "__main__":
+    main()

nexus/mempalace/__init__.py

@@ -13,6 +13,12 @@ from __future__ import annotations
 
 from nexus.mempalace.config import MEMPALACE_PATH, FLEET_WING
 from nexus.mempalace.searcher import search_memories, add_memory, MemPalaceResult
+from nexus.mempalace.conversation_artifacts import (
+    ConversationArtifact,
+    build_request_response_artifact,
+    extract_alexander_request_pairs,
+    normalize_speaker,
+)
 
 __all__ = [
     "MEMPALACE_PATH",
@@ -20,4 +26,8 @@ __all__ = [
     "search_memories",
     "add_memory",
     "MemPalaceResult",
+    "ConversationArtifact",
+    "build_request_response_artifact",
+    "extract_alexander_request_pairs",
+    "normalize_speaker",
 ]

nexus/mempalace/config.py

@@ -40,6 +40,7 @@ CORE_ROOMS: list[str] = [
     "nexus",       # reports, docs, KT
     "issues",      # tickets, backlog
     "experiments", # prototypes, spikes
+    "sovereign",   # Alexander request/response artifacts
 ]
 
 # ── ChromaDB collection name ──────────────────────────────────────────────────

nexus/mempalace/conversation_artifacts.py

@@ -0,0 +1,122 @@
+"""Helpers for preserving Alexander request/response artifacts in MemPalace.
+
+This module provides a small, typed bridge between raw conversation turns and
+MemPalace drawers stored in the shared `sovereign` room. The goal is not to
+solve all future speaker-tagging needs at once; it gives the Nexus one
+canonical artifact shape that other miners and bridges can reuse.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from datetime import datetime, timezone
+from typing import Iterable
+
+_ALEXANDER_ALIASES = {
+    "alexander",
+    "alexander whitestone",
+    "rockachopa",
+    "triptimmy",
+}
+
+
+@dataclass(frozen=True)
+class ConversationArtifact:
+    requester: str
+    responder: str
+    request_text: str
+    response_text: str
+    room: str = "sovereign"
+    timestamp: str = field(default_factory=lambda: datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ"))
+    metadata: dict = field(default_factory=dict)
+
+    @property
+    def text(self) -> str:
+        return (
+            f"# Conversation Artifact\n\n"
+            f"## Alexander Request\n{self.request_text.strip()}\n\n"
+            f"## Wizard Response\n{self.response_text.strip()}\n"
+        )
+
+
+def normalize_speaker(name: str | None) -> str:
+    cleaned = " ".join((name or "").strip().lower().split())
+    if cleaned in _ALEXANDER_ALIASES:
+        return "alexander"
+    return cleaned.replace(" ", "_") or "unknown"
+
+
+def build_request_response_artifact(
+    *,
+    requester: str,
+    responder: str,
+    request_text: str,
+    response_text: str,
+    source: str = "",
+    timestamp: str | None = None,
+    request_timestamp: str | None = None,
+    response_timestamp: str | None = None,
+) -> ConversationArtifact:
+    requester_slug = normalize_speaker(requester)
+    responder_slug = normalize_speaker(responder)
+    ts = timestamp or datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ")
+    metadata = {
+        "artifact_type": "alexander_request_response",
+        "requester": requester_slug,
+        "responder": responder_slug,
+        "speaker_tags": [f"speaker:{requester_slug}", f"speaker:{responder_slug}"],
+        "source": source,
+        "timestamp": ts,
+    }
+    if request_timestamp:
+        metadata["request_timestamp"] = request_timestamp
+    if response_timestamp:
+        metadata["response_timestamp"] = response_timestamp
+    return ConversationArtifact(
+        requester=requester_slug,
+        responder=responder_slug,
+        request_text=request_text,
+        response_text=response_text,
+        timestamp=ts,
+        metadata=metadata,
+    )
+
+
+def extract_alexander_request_pairs(
+    turns: Iterable[dict],
+    *,
+    responder: str,
+    source: str = "",
+) -> list[ConversationArtifact]:
+    responder_slug = normalize_speaker(responder)
+    pending_request: dict | None = None
+    artifacts: list[ConversationArtifact] = []
+
+    for turn in turns:
+        speaker = normalize_speaker(
+            turn.get("speaker") or turn.get("username") or turn.get("author") or turn.get("name")
+        )
+        text = (turn.get("text") or turn.get("content") or "").strip()
+        if not text:
+            continue
+
+        if speaker == "alexander":
+            pending_request = turn
+            continue
+
+        if speaker == responder_slug and pending_request is not None:
+            artifacts.append(
+                build_request_response_artifact(
+                    requester="alexander",
+                    responder=responder_slug,
+                    request_text=(pending_request.get("text") or pending_request.get("content") or "").strip(),
+                    response_text=text,
+                    source=source,
+                    request_timestamp=pending_request.get("timestamp"),
+                    response_timestamp=turn.get("timestamp"),
+                    timestamp=turn.get("timestamp") or pending_request.get("timestamp"),
+                )
+            )
+            pending_request = None
+
+    return artifacts

portal-hot-reload.js

@@ -0,0 +1,105 @@
+/**
+ * Portal Hot-Reload for The Nexus
+ * 
+ * Watches portals.json for changes and hot-reloads portal list
+ * without server restart. Existing connections unaffected.
+ * 
+ * Usage:
+ *   PortalHotReload.start(intervalMs);
+ *   PortalHotReload.stop();
+ *   PortalHotReload.reload(); // manual reload
+ */
+
+const PortalHotReload = (() => {
+  let _interval = null;
+  let _lastHash = '';
+  let _pollInterval = 5000; // 5 seconds
+
+  function _hashPortals(data) {
+    // Simple hash of portal IDs for change detection
+    return data.map(p => p.id || p.name).sort().join(',');
+  }
+
+  async function _checkForChanges() {
+    try {
+      const response = await fetch('./portals.json?t=' + Date.now());
+      if (!response.ok) return;
+      
+      const data = await response.json();
+      const hash = _hashPortals(data);
+      
+      if (hash !== _lastHash) {
+        console.log('[PortalHotReload] Detected change — reloading portals');
+        _lastHash = hash;
+        _reloadPortals(data);
+      }
+    } catch (e) {
+      // Silent fail — file might be mid-write
+    }
+  }
+
+  function _reloadPortals(data) {
+    // Remove old portals from scene
+    if (typeof portals !== 'undefined' && Array.isArray(portals)) {
+      portals.forEach(p => {
+        if (p.group && typeof scene !== 'undefined' && scene) {
+          scene.remove(p.group);
+        }
+      });
+      portals.length = 0;
+    }
+
+    // Create new portals
+    if (typeof createPortals === 'function') {
+      createPortals(data);
+    }
+
+    // Re-register with spatial search if available
+    if (window.SpatialSearch && typeof portals !== 'undefined') {
+      portals.forEach(p => {
+        if (p.config && p.config.name && p.group) {
+          SpatialSearch.register('portal', p, p.config.name);
+        }
+      });
+    }
+
+    // Notify
+    if (typeof addChatMessage === 'function') {
+      addChatMessage('system', `Portals reloaded: ${data.length} portals active`);
+    }
+
+    console.log(`[PortalHotReload] Reloaded ${data.length} portals`);
+  }
+
+  function start(intervalMs) {
+    if (_interval) return;
+    _pollInterval = intervalMs || _pollInterval;
+
+    // Initial load
+    fetch('./portals.json').then(r => r.json()).then(data => {
+      _lastHash = _hashPortals(data);
+    }).catch(() => {});
+
+    _interval = setInterval(_checkForChanges, _pollInterval);
+    console.log(`[PortalHotReload] Watching portals.json every ${_pollInterval}ms`);
+  }
+
+  function stop() {
+    if (_interval) {
+      clearInterval(_interval);
+      _interval = null;
+      console.log('[PortalHotReload] Stopped');
+    }
+  }
+
+  async function reload() {
+    const response = await fetch('./portals.json?t=' + Date.now());
+    const data = await response.json();
+    _lastHash = _hashPortals(data);
+    _reloadPortals(data);
+  }
+
+  return { start, stop, reload };
+})();
+
+window.PortalHotReload = PortalHotReload;

reports/pr-backlog-triage-1471.md

@@ -0,0 +1,119 @@
+# timmy-config PR Backlog Triage — Issue #1471
+
+**Date updated:** 2026-04-21 (Pass 27)
+**Agent:** claude
+**Source issue:** #1471
+
+## Summary
+
+| Metric | Value |
+|--------|-------|
+| PRs when filed | 9 |
+| Peak backlog | 50 |
+| Duplicates closed | 25+ |
+| Dangerous PRs closed | 2+ (#815, #833) |
+| PRs merged (all passes) | 31+ |
+| **Current open PRs** | **0** |
+
+## Pass History
+
+### Pass 1–5 (2026-04-16 to 2026-04-17)
+- Closed 14 duplicate PRs (config templates, shebangs, Makefile fixes, etc.)
+- Closed 9 already-merged PRs (0 unique commits ahead of main)
+- Closed PR #815 (dangerous: claimed Makefile fix, actually deleted 50 files including CI)
+- Created PR #822 as clean replacement for #815
+- Merged/resolved ~20 PRs with add/add conflicts from simultaneous agents
+
+### Pass 6 (2026-04-20)
+- Merged PR #824 — fix: restore pytest collection (7 syntax/import errors)
+- Merged PR #825 — feat: code block normalization tests
+- Merged PR #826 — feat: backfill provenance on all training data
+- Merged PR #830 — feat: training data quality filter
+- Closed PR #831 — .DS_Store committed + 81/82 shebangs already present
+
+### Pass 7 (2026-04-21 ~00:00)
+- Closed PR #831 (duplicate shebangs + .DS_Store committed)
+- Created PR #832 — minimal shebang fix for remaining file + .gitignore
+
+### Pass 8 (2026-04-21 ~00:11)
+- Merged PR #832 (closes #681)
+- Confirmed 0 open PRs
+
+### Pass 9 (2026-04-21 ~00:38)
+- PR #833 appeared: "fix: #596" — claimed crisis response training data
+  - **CLOSED**: contained 30 file deletions (3608 lines), 0 additions
+  - Deleted CI workflows, .gitignore, documentation, training data
+  - Same pattern as PR #815; closed with explanation
+- PR #834 appeared: "feat: stale hermes process cleanup script (#829)"
+  - **MERGED**: adds bin/hermes_cleanup.py + tests/test_hermes_cleanup.py
+  - Clean 2-file addition, mergeable, no conflicts
+- **Confirmed 0 open PRs** after this pass
+
+### Pass 10 (2026-04-21 ~02:00)
+- PR #835 appeared: "feat(#691): training pair provenance tracking — source session + model"
+  - **MERGED**: changes training/training_pair_provenance.py (+91/-3) and training/build_curated.py (+12/-0)
+  - 9 tests pass, adds provenance metadata (session_id, model, timestamp) to training pairs
+  - Closes #691
+- PR #836 appeared: "feat: PR triage automation — categorize, auto-merge safe PRs, file reports (#659)"
+  - **MERGED**: adds scripts/pr-triage.sh (+7), updates scripts/pr_triage.py (+278/-238) and tests/test_pr_triage.py (+152/-128)
+  - 40+ tests, auto-merge capability, org-wide triage, closes #659
+- **Confirmed 0 open PRs** after this pass
+
+### Pass 11 (2026-04-21 ~07:30)
+- PR #837 appeared: "fix: complete all 9 genre scene description files + validation tests (closes #645)"
+  - **MERGED**: adds 154 lines to 1 file — fixes missing `artist`/`timestamp` fields in country genre training data
+  - All 100 country entries now pass schema validation
+- PR #838 appeared: "feat: adversary execution harness for prompt corpora (#652)"
+  - **MERGED**: adds scripts/adversary-harness.py (292 lines) — automated adversary prompt replay, scoring, issue filing
+  - Closes #652
+- PR #839 appeared: "feat: auto-generate scene descriptions from image/video assets (#689)"
+  - **MERGED**: adds scripts/generate_scenes_from_media.py + tests (401 lines, 2 files)
+  - Scans media assets, calls vision model, outputs training pairs with provenance metadata
+  - Closes #689
+- **Confirmed 0 open PRs** after this pass
+
+### Pass 12 (2026-04-21 — final verification)
+- No new PRs since Pass 11
+- Verified via API: **0 open PRs** in timmy-config
+- Issue fully resolved. PR #1625 is mergeable and contains the full audit trail.
+
+### Pass 13–17 (2026-04-21)
+- Repeated verification passes confirmed: **0 open PRs** in timmy-config
+- PR #1625 remains open and mergeable at SHA `55c5be4`
+
+### Pass 18 (2026-04-21 ~12:20)
+- Verified via API: **0 open PRs** in timmy-config
+- No new PRs since Pass 17
+- Issue remains fully resolved. PR #1625 ready to merge.
+
+### Pass 19–27 (2026-04-21)
+- Repeated verification passes confirmed: **0 open PRs** in timmy-config
+- PR #1625 remains open and mergeable (head `c7f79b5`, mergeable=true)
+- No new PRs created since Pass 11 (last action pass)
+
+## Systemic Controls in Place
+
+- `stale-pr-cleanup.py` (fleet-ops PR #301): warns at 3 days, closes at 4 days
+- `pr-capacity.py` (fleet-ops PR #302): max 10 PRs for timmy-config
+- `burn-rotation.py` (fleet-ops PR #297): distributes work across repos
+
+## Pattern: Dangerous Deletion PRs
+
+Multiple PRs have been identified that claim to implement features but actually delete existing infrastructure:
+- PR #815 — claimed Makefile fix, deleted 50 files (closed)
+- PR #833 — claimed crisis response data, deleted 30 files (closed)
+
+**Root cause hypothesis**: Agent generates a PR on a branch accidentally based on an old commit, missing many recent merges. From the agent's perspective those files are "new" on main, making them appear as deletions from its branch.
+
+**Recommendation**: Add a CI check that fails PRs with high deletion-to-addition ratios (e.g., >10 deletions and 0 additions should be flagged for manual review).
+
+## Pre-existing CI Issues (Repo-wide)
+
+These CI checks are failing on `main` and were pre-existing before this triage:
+- YAML Lint
+- Shell Script Lint
+- Python Syntax & Import Check (causes Python Test Suite to be skipped)
+- Smoke Test
+- Architecture Lint / Lint Repository
+
+These are not introduced by any of the merged PRs. Should be addressed in a separate issue.

reports/timmy-config-pr-triage-2026-04-17.md

@@ -0,0 +1,125 @@
+# timmy-config PR Backlog Triage Report
+**Date:** 2026-04-17
+**Issue:** Timmy_Foundation/the-nexus#1471
+**Starting backlog:** 20 open PRs (was 9 when issue was filed)
+
+## Summary of Actions
+
+| Action | Count | PRs |
+|--------|-------|-----|
+| Closed (already merged) | 13 | #802, #804, #805, #807, #808, #809, #810, #811, #812, #813, #814, #816, #817 |
+| Closed (dangerous/wrong) | 1 | #815 |
+| Closed (duplicate) | 4 | #799, #803, #819, #820 |
+| Created (correct fix) | 1 | #822 |
+| **Remaining open** | **2** | #818, #821 |
+
+---
+
+## Closed: Already Merged into Main (13 PRs)
+
+These PRs had 0 unique commits ahead of main — their content was already merged.
+The PRs were left open by an automated system that creates PRs but doesn't close them after merge.
+
+| PR | Title |
+|----|-------|
+| #802 | feat: shared adversary scoring rubric and transcript schema |
+| #804 | fix: hash dedup rotation + bloom filter — bounded memory |
+| #805 | fix: pipeline_state.json daily reset |
+| #807 | test: quality gate test suite |
+| #808 | feat: Token tracker integrated with orchestrator |
+| #809 | fix: training data code block indentation |
+| #810 | feat: PR backlog triage script |
+| #811 | feat: adversary execution harness for prompt corpora |
+| #812 | test: verify training example metadata preservation |
+| #813 | feat: scene data validator tests + CI path fix |
+| #814 | fix: cron fleet audit |
+| #816 | feat: harm facilitation adversary — 200 jailbreak prompts |
+| #817 | feat: quality filter tests |
+
+**Root cause:** Merge workflow merges PRs but doesn't close the PR objects. Or PRs were force-pushed/squash-merged without closing.
+
+---
+
+## Closed: Dangerous PR (1 PR)
+
+### PR #815 — `fix: use PYTHON variable in training Makefile (#660)`
+
+**Status: DANGEROUS — correctly closed without merging.**
+
+This PR claimed to be a simple Makefile fix (add `PYTHON ?= python3` variable) but its actual diff was:
+- **0 files added**
+- **0 files changed**
+- **50 files deleted** — including all `.gitea/workflows/`, `README.md`, `CONTRIBUTING.md`, `GENOME.md`, `HEART.md`, `SOUL.md`, `adversary/` corpus files, and other critical infrastructure
+
+This was a severe agent error — the branch `fix/660` appears to have been created from a different base or the agent accidentally committed a state where those files were missing. **Merging this PR would have destroyed the CI pipeline and core documentation.**
+
+**Fix:** Created PR #822 with the correct, minimal change (only modifies `training/Makefile`).
+
+---
+
+## Closed: Duplicate Training Data PRs (4 PRs)
+
+PRs #799, #803, #819, #820, and #821 all added overlapping training data files. They were created by multiple Claude agents independently implementing the same features without coordination.
+
+**Overlap analysis:**
+
+| File | In main? | #799 | #803 | #819 | #820 | #821 |
+|------|----------|------|------|------|------|------|
+| GENOME.md | YES | ✓ | ✓ | ✓ | ✓ | ✓ |
+| training/data/crisis-response/post-crisis-recovery-500.jsonl | NO | ✓ | - | ✓ | ✓ | ✓ |
+| training/data/prompt-enhancement/dream-descriptions-500.jsonl | NO | - | - | - | - | ✓ |
+| training/data/scene-descriptions/scene-descriptions-country.jsonl | NO | - | - | - | ✓ | ✓ |
+| training/data/scene-descriptions/scene-descriptions-latin.jsonl | NO | - | - | - | ✓ | ✓ |
+| training/provenance.py | NO | - | ✓ | ✓ | ✓ | ✓ |
+
+**Decision:** Kept PR #821 (most complete, includes all scene descriptions + dream-descriptions). Closed #799, #803, #819, #820 as superseded.
+
+---
+
+## Remaining Open PRs (2)
+
+### PR #821 — `feat: 500 dream description prompt enhancement pairs (#602)`
+
+**Status: Needs rebase**
+
+The most complete training data PR. Contains all net-new files. Currently `Mergeable: False` because it conflicts with files already in main (GENOME.md, several training data files that landed in earlier PRs).
+
+**Files NOT yet in main (net-new value):**
+- `training/data/crisis-response/post-crisis-recovery-500.jsonl`
+- `training/data/prompt-enhancement/dream-descriptions-500.jsonl`
+- `training/data/scene-descriptions/scene-descriptions-country.jsonl`
+- `training/data/scene-descriptions/scene-descriptions-hip-hop.jsonl`
+- `training/data/scene-descriptions/scene-descriptions-latin.jsonl`
+- `training/provenance.py`
+- `training/scripts/generate_scene_descriptions.py`
+- `scripts/config_drift_detector.py`
+- `evaluations/adversary/corpora/emotional_manipulation_200.jsonl`
+- `evaluations/adversary/corpora/identity_attacks_200.jsonl`
+
+**Action needed:** Rebase `fix/602` onto current main, keeping only the net-new files.
+
+### PR #818 — `feat: quality gate pipeline validation (#623)`
+
+**Status: Needs rebase**
+
+Adds `bin/quality-gate.py` (+292 lines) and `pipeline/quality_gate.py` (+419 lines) — both are net-new. Currently `Mergeable: False` due to rebase drift.
+
+**Action needed:** Rebase `fix/623` onto current main.
+
+---
+
+## Root Cause Analysis
+
+The PR backlog grew from 9 to 20 during a single day of automated agent activity. The pattern is:
+
+1. **Merge-without-close:** PRs get merged but the PR objects aren't closed, creating phantom open PRs
+2. **Duplicate agent runs:** Multiple agents work the same issue concurrently, producing overlapping PRs
+3. **Wrong-base branches:** Agent PR #815 is a severe example — the agent created a branch from the wrong base, producing a destructive diff
+4. **No coordination signal:** Agents don't check for existing open PRs on the same issue before creating new ones
+
+## Process Recommendations
+
+1. **Auto-close merged PRs:** Add a Gitea webhook or CI step that closes PRs when their head branch is detected in main
+2. **PR dedup check:** Before creating a PR, agents should check `GET /repos/{owner}/{repo}/pulls?state=open&head={branch-prefix}` for existing PRs on the same issue
+3. **Branch safety check:** Before creating a PR, validate that the diff is sane (no massive deletions for a fix PR)
+4. **Issue lock after PR:** Once a PR is created for an issue, lock the issue to prevent other agents from working it simultaneously

reports/timmy-config-pr-triage-2026-04-21.md

@@ -0,0 +1,70 @@
+# timmy-config PR Backlog Triage Report
+**Date:** 2026-04-21
+**Issue:** Timmy_Foundation/the-nexus#1471
+
+## Summary
+
+| Metric | Value |
+|--------|-------|
+| PRs when issue filed | 9 |
+| Peak backlog | 50 |
+| Total passes | 31+ |
+| Duplicates closed | 25+ |
+| Dangerous PRs blocked | 2 (#815, #833) |
+| PRs merged (all passes) | 32+ |
+| **Open PRs now** | **0** |
+
+## Status: RESOLVED
+
+timmy-config PR backlog is fully cleared as of 2026-04-21.
+
+## Pass History
+
+### Pass 1–3 (initial triage)
+- Closed 14 duplicate PRs identified by shared issue refs
+- Backlog grew from 9 → 50 as new agent waves added PRs
+
+### Pass 4–6 (merge wave)
+- Merged 13 cleanly mergeable PRs
+- Resolved 7 add/add conflicts from simultaneous agent PRs
+- Closed dangerous PR #815 (50 file deletions masquerading as a fix)
+
+### Pass 7–8
+- Closed PR #831 (shebang fix with .DS_Store, merge conflicts, 81/82 files already fixed)
+- Created clean replacement PR #832
+- Merged PR #832 (shebang + .gitignore)
+
+### Pass 9–11
+- Closed dangerous PR #833 (30 file deletions, same pattern as #815)
+- Merged PR #834 (stale hermes process cleanup)
+- Merged PR #835 (training pair provenance tracking)
+- Merged PR #836 (PR triage automation with auto-merge)
+- Merged PR #837 (genre scene description files + validation tests)
+- Merged PR #838 (adversary execution harness)
+
+### Pass 12–21 (verification passes)
+- Verified backlog held at 0 across repeated passes
+- No new PRs accumulating
+
+### Pass 30–31
+- Merged PR #840 (JSON schema + validator for scene description training data)
+- Merged PR #842 (MEMORY.md forge domain fix)
+- Confirmed final state: 0 open PRs
+
+## Dangerous PRs Blocked
+
+### PR #815 — "fix: use PYTHON variable in training Makefile"
+- **Actual content:** 50 file deletions (CI workflows, README, GENOME.md, HEART.md, adversary corpus)
+- **Action:** Closed with detailed explanation
+
+### PR #833 — "fix: crisis response training data"
+- **Actual content:** 30 file deletions / 3608 lines removed, 0 additions
+- Files deleted: CI workflows, .gitignore, GENOME.md, CONTRIBUTING.md, training data
+- **Action:** Closed with detailed explanation
+
+## Systemic Tools Created
+
+- `scripts/pr-backlog-triage.py` — identifies duplicate PRs by issue ref
+- `stale-pr-cleanup.py` — warns at 3 days, closes at 4 days
+- `pr-capacity.py` — per-repo PR limits (timmy-config: 10 max)
+- `burn-rotation.py` — rotates work across repos to prevent concentration

server.py

@@ -3,20 +3,37 @@
 The Nexus WebSocket Gateway — Robust broadcast bridge for Timmy's consciousness.
 This server acts as the central hub for the-nexus, connecting the mind (nexus_think.py),
 the body (Evennia/Morrowind), and the visualization surface.
+
+Security features:
+- Binds to 127.0.0.1 by default (localhost only)
+- Optional external binding via NEXUS_WS_HOST environment variable
+- Token-based authentication via NEXUS_WS_TOKEN environment variable
+- Rate limiting on connections
+- Connection logging and monitoring
 """
 import asyncio
 import json
 import logging
+import os
+import pty
 import signal
+import subprocess
 import sys
-from typing import Set
+import time
+from typing import Set, Dict, Optional
+from collections import defaultdict
 
 # Branch protected file - see POLICY.md
 import websockets
 
 # Configuration
-PORT = 8765
-HOST = "0.0.0.0"  # Allow external connections if needed
+PORT      = int(os.environ.get("NEXUS_WS_PORT", "8765"))
+PTY_PORT  = int(os.environ.get("NEXUS_PTY_PORT", "8766"))   # operator shell PTY gateway
+HOST      = os.environ.get("NEXUS_WS_HOST", "127.0.0.1")    # Default to localhost only
+AUTH_TOKEN = os.environ.get("NEXUS_WS_TOKEN", "")           # Empty = no auth required
+RATE_LIMIT_WINDOW = 60  # seconds
+RATE_LIMIT_MAX_CONNECTIONS = 10  # max connections per IP per window
+RATE_LIMIT_MAX_MESSAGES = 100  # max messages per connection per window
 
 # Logging setup
 logging.basicConfig(
@@ -28,15 +45,207 @@ logger = logging.getLogger("nexus-gateway")
 
 # State
 clients: Set[websockets.WebSocketServerProtocol] = set()
+connection_tracker: Dict[str, list] = defaultdict(list)  # IP -> [timestamps]
+message_tracker: Dict[int, list] = defaultdict(list)  # connection_id -> [timestamps]
+
+def check_rate_limit(ip: str) -> bool:
+    """Check if IP has exceeded connection rate limit."""
+    now = time.time()
+    # Clean old entries
+    connection_tracker[ip] = [t for t in connection_tracker[ip] if now - t < RATE_LIMIT_WINDOW]
+    
+    if len(connection_tracker[ip]) >= RATE_LIMIT_MAX_CONNECTIONS:
+        return False
+    
+    connection_tracker[ip].append(now)
+    return True
+
+def check_message_rate_limit(connection_id: int) -> bool:
+    """Check if connection has exceeded message rate limit."""
+    now = time.time()
+    # Clean old entries
+    message_tracker[connection_id] = [t for t in message_tracker[connection_id] if now - t < RATE_LIMIT_WINDOW]
+    
+    if len(message_tracker[connection_id]) >= RATE_LIMIT_MAX_MESSAGES:
+        return False
+    
+    message_tracker[connection_id].append(now)
+    return True
+
+async def authenticate_connection(websocket: websockets.WebSocketServerProtocol) -> bool:
+    """Authenticate WebSocket connection using token."""
+    if not AUTH_TOKEN:
+        # No authentication required
+        return True
+    
+    try:
+        # Wait for authentication message (first message should be auth)
+        auth_message = await asyncio.wait_for(websocket.recv(), timeout=5.0)
+        auth_data = json.loads(auth_message)
+        
+        if auth_data.get("type") != "auth":
+            logger.warning(f"Invalid auth message type from {websocket.remote_address}")
+            return False
+        
+        token = auth_data.get("token", "")
+        if token != AUTH_TOKEN:
+            logger.warning(f"Invalid auth token from {websocket.remote_address}")
+            return False
+        
+        logger.info(f"Authenticated connection from {websocket.remote_address}")
+        return True
+        
+    except asyncio.TimeoutError:
+        logger.warning(f"Authentication timeout from {websocket.remote_address}")
+        return False
+    except json.JSONDecodeError:
+        logger.warning(f"Invalid auth JSON from {websocket.remote_address}")
+        return False
+    except Exception as e:
+        logger.error(f"Authentication error from {websocket.remote_address}: {e}")
+        return False
+
+# ---------------------------------------------------------------------------
+# Git status helper (issue #1695)
+# ---------------------------------------------------------------------------
+def _get_git_status() -> dict:
+    """Return a dict describing the current repo git state."""
+    repo_root = os.path.dirname(os.path.abspath(__file__))
+    try:
+        branch_out = subprocess.check_output(
+            ["git", "rev-parse", "--abbrev-ref", "HEAD"],
+            cwd=repo_root, stderr=subprocess.DEVNULL, text=True
+        ).strip()
+    except Exception:
+        branch_out = "unknown"
+
+    dirty     = False
+    untracked = 0
+    ahead     = 0
+    try:
+        status_out = subprocess.check_output(
+            ["git", "status", "--porcelain", "--branch"],
+            cwd=repo_root, stderr=subprocess.DEVNULL, text=True
+        )
+        for line in status_out.splitlines():
+            if line.startswith("##") and "ahead" in line:
+                import re
+                m = re.search(r"ahead (\d+)", line)
+                if m:
+                    ahead = int(m.group(1))
+            elif line.startswith("??"):
+                untracked += 1
+            elif line and not line.startswith("##"):
+                dirty = True
+    except Exception:
+        pass
+
+    return {
+        "type":      "git_status",
+        "branch":    branch_out,
+        "dirty":     dirty,
+        "untracked": untracked,
+        "ahead":     ahead,
+    }
+
+
+# ---------------------------------------------------------------------------
+# PTY shell handler (issue #1695) — operator cockpit terminal
+# Binds on PTY_PORT (default 8766), localhost only.
+# Each WebSocket connection gets its own PTY subprocess.
+# ---------------------------------------------------------------------------
+async def pty_handler(websocket: websockets.WebSocketServerProtocol):
+    """Spawn a local PTY shell and bridge it to the WebSocket client."""
+    addr = websocket.remote_address
+    logger.info(f"[PTY] Operator shell connection from {addr}")
+
+    shell = os.environ.get("SHELL", "/bin/bash")
+    master_fd, slave_fd = pty.openpty()
+
+    proc = await asyncio.create_subprocess_exec(
+        shell,
+        stdin=slave_fd,
+        stdout=slave_fd,
+        stderr=slave_fd,
+        preexec_fn=os.setsid,
+        close_fds=True,
+    )
+    os.close(slave_fd)
+
+    loop = asyncio.get_running_loop()
+
+    async def pty_to_ws():
+        """Read PTY output and forward to WebSocket."""
+        try:
+            while True:
+                data = await loop.run_in_executor(None, os.read, master_fd, 4096)
+                if not data:
+                    break
+                await websocket.send(data.decode("utf-8", errors="replace"))
+        except (OSError, websockets.exceptions.ConnectionClosed):
+            pass
+
+    async def ws_to_pty():
+        """Read WebSocket input and forward to PTY."""
+        try:
+            async for message in websocket:
+                if isinstance(message, str):
+                    os.write(master_fd, message.encode("utf-8"))
+                else:
+                    os.write(master_fd, message)
+        except (OSError, websockets.exceptions.ConnectionClosed):
+            pass
+
+    reader = asyncio.ensure_future(pty_to_ws())
+    writer = asyncio.ensure_future(ws_to_pty())
+    try:
+        await asyncio.gather(reader, writer)
+    finally:
+        reader.cancel()
+        writer.cancel()
+        try:
+            os.close(master_fd)
+        except OSError:
+            pass
+        try:
+            proc.kill()
+        except ProcessLookupError:
+            pass
+        await proc.wait()
+        logger.info(f"[PTY] Shell session ended for {addr}")
+
 
 async def broadcast_handler(websocket: websockets.WebSocketServerProtocol):
     """Handles individual client connections and message broadcasting."""
-    clients.add(websocket)
     addr = websocket.remote_address
+    ip = addr[0] if addr else "unknown"
+    connection_id = id(websocket)
+    
+    # Check connection rate limit
+    if not check_rate_limit(ip):
+        logger.warning(f"Connection rate limit exceeded for {ip}")
+        await websocket.close(1008, "Rate limit exceeded")
+        return
+    
+    # Authenticate if token is required
+    if not await authenticate_connection(websocket):
+        await websocket.close(1008, "Authentication failed")
+        return
+    
+    clients.add(websocket)
     logger.info(f"Client connected from {addr}. Total clients: {len(clients)}")
     
     try:
         async for message in websocket:
+            # Check message rate limit
+            if not check_message_rate_limit(connection_id):
+                logger.warning(f"Message rate limit exceeded for {addr}")
+                await websocket.send(json.dumps({
+                    "type": "error",
+                    "message": "Message rate limit exceeded"
+                }))
+                continue
+            
             # Parse for logging/validation if it's JSON
             try:
                 data = json.loads(message)
@@ -44,6 +253,11 @@ async def broadcast_handler(websocket: websockets.WebSocketServerProtocol):
                 # Optional: log specific important message types
                 if msg_type in ["agent_register", "thought", "action"]:
                     logger.debug(f"Received {msg_type} from {addr}")
+                # Handle git status requests from the operator cockpit (issue #1695)
+                if msg_type == "git_status_request":
+                    git_info = _get_git_status()
+                    await websocket.send(json.dumps(git_info))
+                    continue
             except (json.JSONDecodeError, TypeError):
                 pass
 
@@ -81,6 +295,20 @@ async def broadcast_handler(websocket: websockets.WebSocketServerProtocol):
 
 async def main():
     """Main server loop with graceful shutdown."""
+    # Log security configuration
+    if AUTH_TOKEN:
+        logger.info("Authentication: ENABLED (token required)")
+    else:
+        logger.warning("Authentication: DISABLED (no token required)")
+    
+    if HOST == "0.0.0.0":
+        logger.warning("Host binding: 0.0.0.0 (all interfaces) - SECURITY RISK")
+    else:
+        logger.info(f"Host binding: {HOST} (localhost only)")
+    
+    logger.info(f"Rate limiting: {RATE_LIMIT_MAX_CONNECTIONS} connections/IP/{RATE_LIMIT_WINDOW}s, "
+                f"{RATE_LIMIT_MAX_MESSAGES} messages/connection/{RATE_LIMIT_WINDOW}s")
+    
     logger.info(f"Starting Nexus WS gateway on ws://{HOST}:{PORT}")
     
     # Set up signal handlers for graceful shutdown
@@ -100,7 +328,10 @@ async def main():
 
     async with websockets.serve(broadcast_handler, HOST, PORT):
         logger.info("Gateway is ready and listening.")
-        await stop
+        # Also start the PTY gateway on PTY_PORT (operator cockpit shell, issue #1695)
+        async with websockets.serve(pty_handler, "127.0.0.1", PTY_PORT):
+            logger.info(f"PTY shell gateway listening on ws://127.0.0.1:{PTY_PORT}/pty")
+            await stop
         
     logger.info("Shutting down Nexus WS gateway...")
     # Close any remaining client connections (handlers may have already cleaned up)

session-manager.js

@@ -0,0 +1,294 @@
+/**
+ * session-manager.js — Session taxonomy primitives for the Nexus operator cockpit
+ *
+ * Operations: create, list, setActive, group, tag, pin, archive, restore, delete
+ * Persistence: localStorage under key `nexus-sessions`
+ *
+ * Refs: issue #1695 — ATLAS cockpit operator patterns
+ * Pattern sources: dodo-reach/hermes-desktop session sidebar, nesquena/hermes-webui session groups
+ */
+
+(function () {
+  'use strict';
+
+  const STORAGE_KEY = 'nexus-sessions';
+  const META_KEY    = 'nexus-sessions-meta';
+
+  // ---------------------------------------------------------------------------
+  // Data model
+  // ---------------------------------------------------------------------------
+  // Session:
+  //   id:        string   (uuid-ish)
+  //   name:      string
+  //   group:     string | null
+  //   tags:      string[]
+  //   pinned:    boolean
+  //   archived:  boolean
+  //   createdAt: number   (epoch ms)
+  //   updatedAt: number
+  //   meta:      {}       (arbitrary caller-supplied data)
+  //
+  // Meta:
+  //   activeId:  string | null
+
+  let _sessions = [];
+  let _activeId = null;
+
+  // ---------------------------------------------------------------------------
+  // Persistence
+  // ---------------------------------------------------------------------------
+  function load() {
+    try {
+      const raw = localStorage.getItem(STORAGE_KEY);
+      _sessions = raw ? JSON.parse(raw) : [];
+    } catch {
+      _sessions = [];
+    }
+    try {
+      const rawMeta = localStorage.getItem(META_KEY);
+      const meta    = rawMeta ? JSON.parse(rawMeta) : {};
+      _activeId     = meta.activeId || null;
+    } catch {
+      _activeId = null;
+    }
+  }
+
+  function save() {
+    try {
+      localStorage.setItem(STORAGE_KEY, JSON.stringify(_sessions));
+      localStorage.setItem(META_KEY,    JSON.stringify({ activeId: _activeId }));
+    } catch (e) {
+      console.warn('[SessionManager] Could not persist sessions:', e);
+    }
+    emit('change', { sessions: _sessions, activeId: _activeId });
+  }
+
+  // ---------------------------------------------------------------------------
+  // ID generation
+  // ---------------------------------------------------------------------------
+  function genId() {
+    return 'sess_' + Date.now().toString(36) + '_' + Math.random().toString(36).slice(2, 7);
+  }
+
+  // ---------------------------------------------------------------------------
+  // Event bus (lightweight)
+  // ---------------------------------------------------------------------------
+  const _listeners = {};
+
+  function on(event, fn) {
+    if (!_listeners[event]) _listeners[event] = [];
+    _listeners[event].push(fn);
+  }
+
+  function off(event, fn) {
+    if (!_listeners[event]) return;
+    _listeners[event] = _listeners[event].filter(f => f !== fn);
+  }
+
+  function emit(event, data) {
+    (_listeners[event] || []).forEach(fn => { try { fn(data); } catch {} });
+  }
+
+  // ---------------------------------------------------------------------------
+  // Core CRUD
+  // ---------------------------------------------------------------------------
+  function create(name, opts = {}) {
+    const session = {
+      id:        genId(),
+      name:      name || 'Untitled Session',
+      group:     opts.group    || null,
+      tags:      opts.tags     || [],
+      pinned:    opts.pinned   || false,
+      archived:  false,
+      createdAt: Date.now(),
+      updatedAt: Date.now(),
+      meta:      opts.meta     || {},
+    };
+    _sessions.push(session);
+    if (!_activeId) _activeId = session.id;
+    save();
+    return session;
+  }
+
+  function get(id) {
+    return _sessions.find(s => s.id === id) || null;
+  }
+
+  function list(opts = {}) {
+    let result = [..._sessions];
+
+    if (opts.group)        result = result.filter(s => s.group === opts.group);
+    if (opts.tag)          result = result.filter(s => s.tags.includes(opts.tag));
+    if (opts.pinned  !== undefined) result = result.filter(s => s.pinned   === opts.pinned);
+    if (opts.archived !== undefined) result = result.filter(s => s.archived === opts.archived);
+
+    // Default: hide archived unless explicitly requested
+    if (opts.archived === undefined) result = result.filter(s => !s.archived);
+
+    // Pinned items first, then by updatedAt desc
+    result.sort((a, b) => {
+      if (a.pinned !== b.pinned) return b.pinned ? 1 : -1;
+      return b.updatedAt - a.updatedAt;
+    });
+
+    return result;
+  }
+
+  function update(id, patch) {
+    const idx = _sessions.findIndex(s => s.id === id);
+    if (idx < 0) return null;
+    _sessions[idx] = { ..._sessions[idx], ...patch, updatedAt: Date.now() };
+    save();
+    return _sessions[idx];
+  }
+
+  function remove(id) {
+    const before = _sessions.length;
+    _sessions = _sessions.filter(s => s.id !== id);
+    if (_activeId === id) _activeId = _sessions.find(s => !s.archived)?.id || null;
+    if (_sessions.length !== before) save();
+  }
+
+  // ---------------------------------------------------------------------------
+  // Taxonomy operations
+  // ---------------------------------------------------------------------------
+
+  /** Set or clear the group for a session. */
+  function group(id, groupName) {
+    return update(id, { group: groupName || null });
+  }
+
+  /** Add one or more tags to a session. */
+  function tag(id, ...tags) {
+    const s = get(id);
+    if (!s) return null;
+    const merged = Array.from(new Set([...s.tags, ...tags.filter(Boolean)]));
+    return update(id, { tags: merged });
+  }
+
+  /** Remove a tag from a session. */
+  function untag(id, tagName) {
+    const s = get(id);
+    if (!s) return null;
+    return update(id, { tags: s.tags.filter(t => t !== tagName) });
+  }
+
+  /** Toggle pin state. */
+  function pin(id) {
+    const s = get(id);
+    if (!s) return null;
+    return update(id, { pinned: !s.pinned });
+  }
+
+  /** Archive a session (hides from default list). */
+  function archive(id) {
+    return update(id, { archived: true, pinned: false });
+  }
+
+  /** Restore an archived session. */
+  function restore(id) {
+    return update(id, { archived: false });
+  }
+
+  // ---------------------------------------------------------------------------
+  // Active session
+  // ---------------------------------------------------------------------------
+  function getActive() {
+    return _activeId;
+  }
+
+  function getActiveSession() {
+    return _activeId ? get(_activeId) : null;
+  }
+
+  function setActive(id) {
+    if (!get(id)) return false;
+    _activeId = id;
+    save();
+    return true;
+  }
+
+  // ---------------------------------------------------------------------------
+  // Group helpers
+  // ---------------------------------------------------------------------------
+  function listGroups() {
+    const seen = new Set();
+    _sessions.forEach(s => { if (s.group) seen.add(s.group); });
+    return Array.from(seen).sort();
+  }
+
+  function listTags() {
+    const seen = new Set();
+    _sessions.forEach(s => s.tags.forEach(t => seen.add(t)));
+    return Array.from(seen).sort();
+  }
+
+  // ---------------------------------------------------------------------------
+  // Import / export (for backup / hand-off)
+  // ---------------------------------------------------------------------------
+  function exportAll() {
+    return JSON.stringify({ sessions: _sessions, activeId: _activeId }, null, 2);
+  }
+
+  function importAll(json) {
+    try {
+      const data = JSON.parse(json);
+      if (!Array.isArray(data.sessions)) throw new Error('Invalid format');
+      _sessions = data.sessions;
+      _activeId = data.activeId || null;
+      save();
+      return true;
+    } catch (e) {
+      console.error('[SessionManager] Import failed:', e);
+      return false;
+    }
+  }
+
+  // ---------------------------------------------------------------------------
+  // Init
+  // ---------------------------------------------------------------------------
+  load();
+
+  // Create a default session if store is empty
+  if (_sessions.length === 0) {
+    create('Default', { tags: ['auto'], meta: { auto: true } });
+  }
+
+  // ---------------------------------------------------------------------------
+  // Public API
+  // ---------------------------------------------------------------------------
+  window.SessionManager = {
+    create,
+    get,
+    list,
+    update,
+    remove,
+
+    // Taxonomy
+    group,
+    tag,
+    untag,
+    pin,
+    archive,
+    restore,
+
+    // Active session
+    getActive,
+    getActiveSession,
+    setActive,
+
+    // Helpers
+    listGroups,
+    listTags,
+
+    // Import/export
+    exportAll,
+    importAll,
+
+    // Events
+    on,
+    off,
+  };
+
+  console.info('[SessionManager] Loaded. Sessions:', _sessions.length, '| Active:', _activeId);
+})();

style.css

@@ -2928,9 +2928,309 @@ body.operator-mode #mode-label {
   .reasoning-trace {
     width: 280px;
   }
-  
+
   .trace-content {
     max-height: 200px;
   }
 }
 
+/* ==========================================================================
+   Operator Inspector Rail — issue #1695
+   ========================================================================== */
+
+.cockpit-inspector {
+  position: fixed;
+  right: 0;
+  top: 0;
+  bottom: 0;
+  width: 280px;
+  background: rgba(5, 8, 20, 0.94);
+  border-left: 1px solid rgba(74, 240, 192, 0.18);
+  display: flex;
+  flex-direction: column;
+  z-index: 1200;
+  font-family: 'JetBrains Mono', 'Courier New', monospace;
+  font-size: 11px;
+  color: #c8e8ff;
+  backdrop-filter: blur(8px);
+  transition: transform 0.25s ease, width 0.25s ease;
+  overflow: hidden;
+}
+
+.cockpit-inspector.collapsed {
+  width: 32px;
+}
+
+.cockpit-inspector.collapsed .ci-header,
+.cockpit-inspector.collapsed .ci-body {
+  display: none;
+}
+
+/* Toggle button */
+.ci-toggle-btn {
+  position: absolute;
+  left: -22px;
+  top: 50%;
+  transform: translateY(-50%);
+  width: 22px;
+  height: 44px;
+  background: rgba(5, 8, 20, 0.9);
+  border: 1px solid rgba(74, 240, 192, 0.25);
+  border-right: none;
+  color: #4af0c0;
+  cursor: pointer;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  font-size: 10px;
+  border-radius: 4px 0 0 4px;
+  z-index: 1;
+}
+
+.ci-toggle-btn:hover { background: rgba(74, 240, 192, 0.1); }
+.ci-toggle-icon { line-height: 1; }
+
+/* Header */
+.ci-header {
+  display: flex;
+  align-items: center;
+  gap: 6px;
+  padding: 10px 10px 8px;
+  border-bottom: 1px solid rgba(74, 240, 192, 0.12);
+  background: rgba(74, 240, 192, 0.04);
+  flex-shrink: 0;
+}
+
+.ci-header-icon { color: #4af0c0; font-size: 13px; }
+.ci-header-title {
+  flex: 1;
+  font-size: 10px;
+  font-weight: 600;
+  letter-spacing: 0.12em;
+  color: #4af0c0;
+  text-transform: uppercase;
+}
+
+.ci-header-actions { display: flex; gap: 4px; }
+
+/* Generic icon button */
+.ci-icon-btn {
+  background: none;
+  border: 1px solid rgba(74, 240, 192, 0.2);
+  color: #7ab8d8;
+  font-size: 11px;
+  padding: 2px 5px;
+  border-radius: 3px;
+  cursor: pointer;
+  font-family: inherit;
+  line-height: 1.4;
+}
+.ci-icon-btn:hover { border-color: #4af0c0; color: #4af0c0; }
+
+/* Body scroll area */
+.ci-body {
+  flex: 1;
+  overflow-y: auto;
+  overflow-x: hidden;
+  scrollbar-width: thin;
+  scrollbar-color: rgba(74,240,192,0.2) transparent;
+}
+
+/* Section */
+.ci-section {
+  border-bottom: 1px solid rgba(74, 240, 192, 0.08);
+}
+
+.ci-section-header {
+  display: flex;
+  align-items: center;
+  gap: 5px;
+  padding: 7px 10px 6px;
+  font-size: 9px;
+  font-weight: 700;
+  letter-spacing: 0.14em;
+  color: #7ab8d8;
+  text-transform: uppercase;
+  cursor: pointer;
+  user-select: none;
+  background: rgba(255,255,255,0.02);
+}
+
+.ci-section-header:hover { color: #4af0c0; }
+.ci-section-icon { color: #4af0c0; font-size: 11px; width: 14px; text-align: center; }
+.ci-section-actions { margin-left: auto; display: flex; gap: 4px; }
+
+.ci-section-badge {
+  margin-left: auto;
+  font-size: 9px;
+  padding: 1px 5px;
+  background: rgba(74,240,192,0.08);
+  border-radius: 8px;
+  color: #4af0c0;
+  min-width: 18px;
+  text-align: center;
+}
+
+.ci-section-badge.badge-warn { background: rgba(255,160,40,0.15); color: #ffa028; }
+.ci-section-badge.badge-ok   { background: rgba(74,240,192,0.12); color: #4af0c0; }
+
+.ci-section-body { padding: 6px 10px 8px; }
+
+.ci-empty-hint {
+  color: rgba(200,232,255,0.35);
+  font-size: 10px;
+  font-style: italic;
+  padding: 2px 0;
+}
+
+/* Tag chip */
+.ci-tag {
+  display: inline-block;
+  padding: 1px 5px;
+  background: rgba(74,240,192,0.08);
+  border: 1px solid rgba(74,240,192,0.18);
+  border-radius: 3px;
+  font-size: 9px;
+  color: #7ab8d8;
+  margin-right: 2px;
+}
+
+.tag-pin     { border-color: rgba(255,180,40,0.4); background: rgba(255,180,40,0.08); }
+.tag-archive { border-color: rgba(150,150,200,0.3); background: rgba(150,150,200,0.06); }
+
+/* Git section rows */
+.ci-git-row {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  padding: 2px 0;
+  font-size: 10px;
+}
+.ci-git-label { color: rgba(200,232,255,0.5); }
+.ci-git-value { color: #c8e8ff; font-weight: 500; }
+.ci-dirty  { color: #ffa028; }
+.ci-clean  { color: #4af0c0; }
+
+/* Agent health rows */
+.ci-agent-row {
+  display: flex;
+  align-items: flex-start;
+  gap: 7px;
+  padding: 5px 0;
+  border-bottom: 1px solid rgba(255,255,255,0.04);
+}
+.ci-agent-row:last-child { border-bottom: none; }
+
+.ci-agent-dot {
+  width: 7px; height: 7px;
+  border-radius: 50%;
+  margin-top: 4px;
+  flex-shrink: 0;
+}
+.agent-idle    { background: #4af0c0; }
+.agent-working { background: #ffa028; box-shadow: 0 0 4px #ffa028; animation: agent-pulse 1s infinite; }
+.agent-error   { background: #ff4060; }
+.agent-unknown { background: rgba(200,232,255,0.3); }
+
+@keyframes agent-pulse {
+  0%, 100% { opacity: 1; }
+  50%       { opacity: 0.4; }
+}
+
+.ci-agent-info  { flex: 1; min-width: 0; }
+.ci-agent-id    { font-size: 10px; color: #c8e8ff; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; }
+.ci-agent-meta  { display: flex; flex-wrap: wrap; gap: 3px; margin-top: 2px; }
+.ci-agent-task  { font-size: 9px; color: rgba(200,232,255,0.5); white-space: nowrap; overflow: hidden; text-overflow: ellipsis; max-width: 140px; }
+
+.ci-agent-status-label {
+  font-size: 9px;
+  padding: 1px 5px;
+  border-radius: 3px;
+  border: 1px solid transparent;
+  white-space: nowrap;
+  flex-shrink: 0;
+}
+.ci-agent-status-label.agent-idle    { border-color: rgba(74,240,192,0.3);  color: #4af0c0; }
+.ci-agent-status-label.agent-working { border-color: rgba(255,160,40,0.3);  color: #ffa028; }
+.ci-agent-status-label.agent-error   { border-color: rgba(255,64,96,0.3);   color: #ff4060; }
+.ci-agent-status-label.agent-unknown { border-color: rgba(200,232,255,0.2); color: rgba(200,232,255,0.5); }
+
+/* Session rows */
+.ci-session-row {
+  display: flex;
+  align-items: center;
+  gap: 6px;
+  padding: 5px 6px;
+  border-radius: 4px;
+  cursor: pointer;
+  margin-bottom: 2px;
+}
+.ci-session-row:hover { background: rgba(74,240,192,0.06); }
+.ci-session-row.active { background: rgba(74,240,192,0.1); border-left: 2px solid #4af0c0; }
+
+.ci-session-info  { flex: 1; min-width: 0; }
+.ci-session-name  { font-size: 10px; color: #c8e8ff; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; }
+.ci-session-meta  { display: flex; flex-wrap: wrap; gap: 3px; margin-top: 2px; }
+.ci-session-actions { display: flex; gap: 2px; }
+
+/* Artifact rows */
+.ci-artifact-row {
+  display: flex;
+  align-items: center;
+  gap: 7px;
+  padding: 4px 0;
+  border-bottom: 1px solid rgba(255,255,255,0.04);
+}
+.ci-artifact-row:last-child { border-bottom: none; }
+.ci-artifact-icon { font-size: 13px; flex-shrink: 0; }
+.ci-artifact-info { flex: 1; min-width: 0; }
+.ci-artifact-name { font-size: 10px; color: #c8e8ff; white-space: nowrap; overflow: hidden; text-overflow: ellipsis; }
+.ci-artifact-path { font-size: 9px; color: rgba(200,232,255,0.4); white-space: nowrap; overflow: hidden; text-overflow: ellipsis; }
+.ci-artifact-type { flex-shrink: 0; }
+
+/* Memory / skills rows */
+.ci-mem-row {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  padding: 3px 0;
+}
+.ci-mem-region { font-size: 10px; color: #c8e8ff; }
+
+/* Terminal section */
+.ci-terminal-section { flex: 0 0 auto; }
+
+.ci-terminal-status {
+  display: flex;
+  align-items: center;
+  gap: 6px;
+  padding: 4px 10px;
+  font-size: 10px;
+  color: rgba(200,232,255,0.5);
+}
+
+.ci-term-dot {
+  width: 6px; height: 6px;
+  border-radius: 50%;
+  flex-shrink: 0;
+}
+.ci-term-dot.disconnected { background: rgba(200,232,255,0.25); }
+.ci-term-dot.connecting   { background: #ffa028; animation: agent-pulse 0.8s infinite; }
+.ci-term-dot.connected    { background: #4af0c0; }
+
+.ci-terminal-mount {
+  margin: 0 10px 8px;
+  border: 1px solid rgba(74,240,192,0.15);
+  border-radius: 3px;
+  overflow: hidden;
+  background: #0a0e1a;
+}
+
+/* Ensure inspector doesn't cover up the 3D canvas on small screens */
+@media (max-width: 900px) {
+  .cockpit-inspector { width: 220px; }
+}
+@media (max-width: 600px) {
+  .cockpit-inspector { display: none; }
+}
+

tests/load/websocket_load_test.py

@@ -0,0 +1,193 @@
+#!/usr/bin/env python3
+"""
+WebSocket Load Test — Benchmark concurrent user sessions on the Nexus gateway.
+
+Tests:
+- Concurrent WebSocket connections
+- Message throughput under load
+- Memory profiling per connection
+- Connection failure/recovery
+
+Usage:
+    python3 tests/load/websocket_load_test.py                    # default (50 users)
+    python3 tests/load/websocket_load_test.py --users 200        # 200 concurrent
+    python3 tests/load/websocket_load_test.py --duration 60      # 60 second test
+    python3 tests/load/websocket_load_test.py --json             # JSON output
+
+Ref: #1505
+"""
+
+import asyncio
+import json
+import os
+import sys
+import time
+import argparse
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+WS_URL = os.environ.get("WS_URL", "ws://localhost:8765")
+
+
+@dataclass
+class ConnectionStats:
+    connected: bool = False
+    connect_time_ms: float = 0
+    messages_sent: int = 0
+    messages_received: int = 0
+    errors: int = 0
+    latencies: List[float] = field(default_factory=list)
+    disconnected: bool = False
+
+
+async def ws_client(user_id: int, duration: int, stats: ConnectionStats, ws_url: str = WS_URL):
+    """Single WebSocket client for load testing."""
+    try:
+        import websockets
+    except ImportError:
+        # Fallback: use raw asyncio
+        stats.errors += 1
+        return
+
+    try:
+        start = time.time()
+        async with websockets.connect(ws_url, open_timeout=5) as ws:
+            stats.connect_time_ms = (time.time() - start) * 1000
+            stats.connected = True
+
+            # Send periodic messages for the duration
+            end_time = time.time() + duration
+            msg_count = 0
+            while time.time() < end_time:
+                try:
+                    msg_start = time.time()
+                    message = json.dumps({
+                        "type": "chat",
+                        "user": f"load-test-{user_id}",
+                        "content": f"Load test message {msg_count} from user {user_id}",
+                    })
+                    await ws.send(message)
+                    stats.messages_sent += 1
+
+                    # Wait for response (with timeout)
+                    try:
+                        response = await asyncio.wait_for(ws.recv(), timeout=5.0)
+                        stats.messages_received += 1
+                        latency = (time.time() - msg_start) * 1000
+                        stats.latencies.append(latency)
+                    except asyncio.TimeoutError:
+                        stats.errors += 1
+
+                    msg_count += 1
+                    await asyncio.sleep(0.5)  # 2 messages/sec per user
+
+                except websockets.exceptions.ConnectionClosed:
+                    stats.disconnected = True
+                    break
+                except Exception:
+                    stats.errors += 1
+
+    except Exception as e:
+        stats.errors += 1
+        if "Connection refused" in str(e) or "connect" in str(e).lower():
+            pass  # Expected if server not running
+
+
+async def run_load_test(users: int, duration: int, ws_url: str = WS_URL) -> dict:
+    """Run the load test with N concurrent users."""
+    stats = [ConnectionStats() for _ in range(users)]
+
+    print(f"  Starting {users} concurrent connections for {duration}s...")
+    start = time.time()
+
+    tasks = [ws_client(i, duration, stats[i], ws_url) for i in range(users)]
+    await asyncio.gather(*tasks, return_exceptions=True)
+
+    total_time = time.time() - start
+
+    # Aggregate results
+    connected = sum(1 for s in stats if s.connected)
+    total_sent = sum(s.messages_sent for s in stats)
+    total_received = sum(s.messages_received for s in stats)
+    total_errors = sum(s.errors for s in stats)
+    disconnected = sum(1 for s in stats if s.disconnected)
+
+    all_latencies = []
+    for s in stats:
+        all_latencies.extend(s.latencies)
+
+    avg_latency = sum(all_latencies) / len(all_latencies) if all_latencies else 0
+    p95_latency = sorted(all_latencies)[int(len(all_latencies) * 0.95)] if all_latencies else 0
+    p99_latency = sorted(all_latencies)[int(len(all_latencies) * 0.99)] if all_latencies else 0
+
+    avg_connect_time = sum(s.connect_time_ms for s in stats if s.connected) / connected if connected else 0
+
+    return {
+        "users": users,
+        "duration_seconds": round(total_time, 1),
+        "connected": connected,
+        "connect_rate": round(connected / users * 100, 1),
+        "messages_sent": total_sent,
+        "messages_received": total_received,
+        "throughput_msg_per_sec": round(total_sent / total_time, 1) if total_time > 0 else 0,
+        "avg_latency_ms": round(avg_latency, 1),
+        "p95_latency_ms": round(p95_latency, 1),
+        "p99_latency_ms": round(p99_latency, 1),
+        "avg_connect_time_ms": round(avg_connect_time, 1),
+        "errors": total_errors,
+        "disconnected": disconnected,
+    }
+
+
+def print_report(result: dict):
+    """Print load test report."""
+    print(f"\n{'='*60}")
+    print(f"  WEBSOCKET LOAD TEST REPORT")
+    print(f"{'='*60}\n")
+
+    print(f"  Connections:    {result['connected']}/{result['users']} ({result['connect_rate']}%)")
+    print(f"  Duration:       {result['duration_seconds']}s")
+    print(f"  Messages sent:  {result['messages_sent']}")
+    print(f"  Messages recv:  {result['messages_received']}")
+    print(f"  Throughput:     {result['throughput_msg_per_sec']} msg/s")
+    print(f"  Avg connect:    {result['avg_connect_time_ms']}ms")
+    print()
+    print(f"  Latency:")
+    print(f"    Avg: {result['avg_latency_ms']}ms")
+    print(f"    P95: {result['p95_latency_ms']}ms")
+    print(f"    P99: {result['p99_latency_ms']}ms")
+    print()
+    print(f"  Errors:         {result['errors']}")
+    print(f"  Disconnected:   {result['disconnected']}")
+
+    # Verdict
+    if result['connect_rate'] >= 95 and result['errors'] == 0:
+        print(f"\n  ✅ PASS")
+    elif result['connect_rate'] >= 80:
+        print(f"\n  ⚠️  DEGRADED")
+    else:
+        print(f"\n  ❌ FAIL")
+
+
+def main():
+    parser = argparse.ArgumentParser(description="WebSocket Load Test")
+    parser.add_argument("--users", type=int, default=50, help="Concurrent users")
+    parser.add_argument("--duration", type=int, default=30, help="Test duration in seconds")
+    parser.add_argument("--json", action="store_true", help="JSON output")
+    parser.add_argument("--url", default=WS_URL, help="WebSocket URL")
+    args = parser.parse_args()
+
+    ws_url = args.url
+
+    print(f"\nWebSocket Load Test — {args.users} users, {args.duration}s\n")
+
+    result = asyncio.run(run_load_test(args.users, args.duration, ws_url))
+
+    if args.json:
+        print(json.dumps(result, indent=2))
+    else:
+        print_report(result)
+
+
+if __name__ == "__main__":
+    main()

tests/test_agent_memory.py

@@ -20,6 +20,7 @@ from agent.memory import (
     SessionTranscript,
     create_agent_memory,
 )
+from nexus.mempalace.conversation_artifacts import ConversationArtifact
 from agent.memory_hooks import MemoryHooks
 
 
@@ -184,6 +185,24 @@ class TestAgentMemory:
         doc_id = mem.write_diary()
         assert doc_id is None  # MemPalace unavailable
 
+    def test_remember_alexander_request_response_uses_sovereign_room(self):
+        mem = AgentMemory(agent_name="allegro")
+        mem._available = True
+        with patch("nexus.mempalace.searcher.add_memory", return_value="doc-123") as add_memory:
+            doc_id = mem.remember_alexander_request_response(
+                request_text="Catalog my requests.",
+                response_text="I will preserve them as artifacts.",
+                requester="Alexander Whitestone",
+                source="telegram:timmy-time",
+            )
+
+        assert doc_id == "doc-123"
+        kwargs = add_memory.call_args.kwargs
+        assert kwargs["room"] == "sovereign"
+        assert kwargs["wing"] == mem.wing
+        assert kwargs["extra_metadata"]["artifact_type"] == "alexander_request_response"
+        assert kwargs["extra_metadata"]["speaker_tags"] == ["speaker:alexander", "speaker:allegro"]
+
 
 # ---------------------------------------------------------------------------
 # MemoryHooks tests

tests/test_chronicle.py

@@ -0,0 +1,211 @@
+"""
+Tests for nexus.chronicle — emergent narrative from agent interactions.
+"""
+
+import json
+import time
+from pathlib import Path
+
+import pytest
+
+from nexus.chronicle import (
+    AgentEvent,
+    ChronicleWriter,
+    EventKind,
+    event_from_commit,
+    event_from_gitea_issue,
+    event_from_heartbeat,
+)
+
+
+# ---------------------------------------------------------------------------
+# AgentEvent
+# ---------------------------------------------------------------------------
+
+class TestAgentEvent:
+    def test_roundtrip(self):
+        evt = AgentEvent(
+            kind=EventKind.DISPATCH,
+            agent="claude",
+            detail="took issue #42",
+        )
+        assert AgentEvent.from_dict(evt.to_dict()).kind == EventKind.DISPATCH
+        assert AgentEvent.from_dict(evt.to_dict()).agent == "claude"
+        assert AgentEvent.from_dict(evt.to_dict()).detail == "took issue #42"
+
+    def test_default_timestamp_is_recent(self):
+        before = time.time()
+        evt = AgentEvent(kind=EventKind.IDLE, agent="mimo")
+        after = time.time()
+        assert before <= evt.timestamp <= after
+
+    def test_all_event_kinds_are_valid_strings(self):
+        for kind in EventKind:
+            evt = AgentEvent(kind=kind, agent="test-agent")
+            d = evt.to_dict()
+            assert d["kind"] == kind.value
+            restored = AgentEvent.from_dict(d)
+            assert restored.kind == kind
+
+
+# ---------------------------------------------------------------------------
+# ChronicleWriter — basic ingestion and render
+# ---------------------------------------------------------------------------
+
+class TestChronicleWriter:
+    @pytest.fixture
+    def writer(self, tmp_path):
+        return ChronicleWriter(log_path=tmp_path / "chronicle.jsonl")
+
+    def test_empty_render(self, writer):
+        text = writer.render()
+        assert "empty" in text.lower()
+
+    def test_single_event_render(self, writer):
+        writer.ingest(AgentEvent(kind=EventKind.DISPATCH, agent="claude", detail="issue #1"))
+        text = writer.render()
+        assert "claude" in text
+        assert "issue #1" in text
+
+    def test_render_covers_timestamps(self, writer):
+        writer.ingest(AgentEvent(kind=EventKind.DISPATCH, agent="a", detail="start"))
+        writer.ingest(AgentEvent(kind=EventKind.COMMIT, agent="a", detail="done"))
+        text = writer.render()
+        assert "chronicle covers" in text.lower()
+
+    def test_events_persisted_to_disk(self, writer, tmp_path):
+        writer.ingest(AgentEvent(kind=EventKind.COMMIT, agent="claude", detail="feat: x"))
+        lines = (tmp_path / "chronicle.jsonl").read_text().strip().splitlines()
+        assert len(lines) == 1
+        data = json.loads(lines[0])
+        assert data["kind"] == "commit"
+        assert data["agent"] == "claude"
+
+    def test_load_existing_on_init(self, tmp_path):
+        log = tmp_path / "chronicle.jsonl"
+        evt = AgentEvent(kind=EventKind.PUSH, agent="mimo", detail="pushed branch")
+        log.write_text(json.dumps(evt.to_dict()) + "\n")
+
+        writer2 = ChronicleWriter(log_path=log)
+        assert len(writer2._events) == 1
+        assert writer2._events[0].kind == EventKind.PUSH
+
+    def test_malformed_lines_are_skipped(self, tmp_path):
+        log = tmp_path / "chronicle.jsonl"
+        log.write_text("not-json\n{}\n")
+        # Should not raise
+        writer2 = ChronicleWriter(log_path=log)
+        assert writer2._events == []
+
+    def test_template_rotation(self, writer):
+        """Consecutive events of the same kind use different templates."""
+        sentences = set()
+        for _ in range(3):
+            writer.ingest(AgentEvent(kind=EventKind.HEARTBEAT, agent="claude"))
+        text = writer.render()
+        # At least one of the template variants should appear
+        assert "pulse" in text or "breathed" in text or "checked in" in text
+
+    def test_render_markdown(self, writer):
+        writer.ingest(AgentEvent(kind=EventKind.PR_OPEN, agent="claude", detail="PR #99"))
+        md = writer.render_markdown()
+        assert md.startswith("# Chronicle")
+        assert "PR #99" in md
+
+    def test_summary(self, writer):
+        writer.ingest(AgentEvent(kind=EventKind.DISPATCH, agent="claude", detail="x"))
+        writer.ingest(AgentEvent(kind=EventKind.COMMIT, agent="claude", detail="y"))
+        s = writer.summary()
+        assert s["total_events"] == 2
+        assert "claude" in s["agents"]
+        assert s["kind_counts"]["dispatch"] == 1
+        assert s["kind_counts"]["commit"] == 1
+
+    def test_max_events_limit(self, writer):
+        for i in range(10):
+            writer.ingest(AgentEvent(kind=EventKind.IDLE, agent="a", detail=str(i)))
+        text = writer.render(max_events=3)
+        # Only 3 events should appear in prose — check coverage header
+        assert "3 event(s)" in text
+
+
+# ---------------------------------------------------------------------------
+# Arc detection
+# ---------------------------------------------------------------------------
+
+class TestArcDetection:
+    @pytest.fixture
+    def writer(self, tmp_path):
+        return ChronicleWriter(log_path=tmp_path / "chronicle.jsonl")
+
+    def _ingest(self, writer, *kinds, agent="claude"):
+        for k in kinds:
+            writer.ingest(AgentEvent(kind=k, agent=agent, detail="x"))
+
+    def test_struggle_and_recovery_arc(self, writer):
+        self._ingest(writer, EventKind.DISPATCH, EventKind.ERROR, EventKind.RECOVERY)
+        text = writer.render()
+        assert "struggle" in text.lower() or "trouble" in text.lower()
+
+    def test_no_arc_when_no_pattern(self, writer):
+        self._ingest(writer, EventKind.IDLE)
+        text = writer.render()
+        # Should not include arc language (only 1 event, no pattern)
+        assert "converged" not in text
+        assert "struggle" not in text
+
+    def test_solo_sprint_arc(self, writer):
+        self._ingest(
+            writer,
+            EventKind.DISPATCH,
+            EventKind.COMMIT,
+            EventKind.PR_OPEN,
+            EventKind.PR_MERGE,
+        )
+        text = writer.render()
+        assert "solo" in text.lower() or "alone" in text.lower()
+
+    def test_fleet_convergence_arc(self, writer, tmp_path):
+        writer2 = ChronicleWriter(log_path=tmp_path / "chronicle.jsonl")
+        writer2.ingest(AgentEvent(kind=EventKind.DISPATCH, agent="claude", detail="x"))
+        writer2.ingest(AgentEvent(kind=EventKind.COLLABORATION, agent="mimo", detail="x"))
+        writer2.ingest(AgentEvent(kind=EventKind.COMMIT, agent="claude", detail="x"))
+        text = writer2.render()
+        assert "converged" in text.lower() or "fleet" in text.lower()
+
+    def test_silent_grind_arc(self, writer):
+        self._ingest(writer, EventKind.COMMIT, EventKind.COMMIT, EventKind.COMMIT)
+        text = writer.render()
+        assert "steady" in text.lower() or "quiet" in text.lower() or "grind" in text.lower()
+
+    def test_abandon_then_retry_arc(self, writer):
+        self._ingest(writer, EventKind.DISPATCH, EventKind.ABANDON, EventKind.DISPATCH)
+        text = writer.render()
+        assert "let go" in text.lower() or "abandon" in text.lower() or "called again" in text.lower()
+
+
+# ---------------------------------------------------------------------------
+# Convenience constructors
+# ---------------------------------------------------------------------------
+
+class TestConvenienceConstructors:
+    def test_event_from_gitea_issue(self):
+        payload = {"number": 42, "title": "feat: add narrative engine"}
+        evt = event_from_gitea_issue(payload, agent="claude")
+        assert evt.kind == EventKind.DISPATCH
+        assert "42" in evt.detail
+        assert evt.agent == "claude"
+
+    def test_event_from_heartbeat(self):
+        hb = {"model": "claude-sonnet", "status": "thinking", "cycle": 7}
+        evt = event_from_heartbeat(hb)
+        assert evt.kind == EventKind.HEARTBEAT
+        assert evt.agent == "claude-sonnet"
+        assert "7" in evt.detail
+
+    def test_event_from_commit(self):
+        commit = {"message": "feat: chronicle\n\nFixes #1607", "sha": "abc1234567"}
+        evt = event_from_commit(commit, agent="claude")
+        assert evt.kind == EventKind.COMMIT
+        assert evt.detail == "feat: chronicle"  # subject line only
+        assert evt.metadata["sha"] == "abc12345"

tests/test_conversation_artifacts.py

@@ -0,0 +1,58 @@
+from pathlib import Path
+
+import yaml
+
+from nexus.mempalace.config import CORE_ROOMS
+from nexus.mempalace.conversation_artifacts import (
+    ConversationArtifact,
+    build_request_response_artifact,
+    extract_alexander_request_pairs,
+    normalize_speaker,
+)
+
+
+def test_sovereign_room_is_core_room() -> None:
+    assert "sovereign" in CORE_ROOMS
+    rooms_yaml = yaml.safe_load(Path("mempalace/rooms.yaml").read_text())
+    assert any(room["key"] == "sovereign" for room in rooms_yaml["core_rooms"])
+
+
+def test_normalize_speaker_maps_alexander_variants() -> None:
+    assert normalize_speaker("Alexander Whitestone") == "alexander"
+    assert normalize_speaker("Rockachopa") == "alexander"
+    assert normalize_speaker("  ALEXANDER  ") == "alexander"
+    assert normalize_speaker("Bezalel") == "bezalel"
+
+
+def test_build_request_response_artifact_creates_sovereign_metadata() -> None:
+    artifact = build_request_response_artifact(
+        requester="Alexander Whitestone",
+        responder="Allegro",
+        request_text="Please organize my conversation artifacts.",
+        response_text="I will catalog them under a sovereign room.",
+        source="telegram:timmy-time",
+        timestamp="2026-04-16T01:30:00Z",
+    )
+
+    assert isinstance(artifact, ConversationArtifact)
+    assert artifact.room == "sovereign"
+    assert artifact.metadata["speaker_tags"] == ["speaker:alexander", "speaker:allegro"]
+    assert artifact.metadata["artifact_type"] == "alexander_request_response"
+    assert artifact.metadata["responder"] == "allegro"
+    assert "## Alexander Request" in artifact.text
+    assert "## Wizard Response" in artifact.text
+
+
+def test_extract_alexander_request_pairs_finds_following_wizard_response() -> None:
+    turns = [
+        {"speaker": "Alexander Whitestone", "text": "Catalog my requests as artifacts.", "timestamp": "2026-04-16T01:00:00Z"},
+        {"speaker": "Allegro", "text": "I'll build a sovereign room contract.", "timestamp": "2026-04-16T01:01:00Z"},
+        {"speaker": "Alexander", "text": "Make sure my words are easy to recall.", "timestamp": "2026-04-16T01:02:00Z"},
+        {"speaker": "Allegro", "text": "I will tag them with speaker metadata.", "timestamp": "2026-04-16T01:03:00Z"},
+    ]
+
+    artifacts = extract_alexander_request_pairs(turns, responder="Allegro", source="telegram")
+
+    assert len(artifacts) == 2
+    assert artifacts[0].metadata["request_timestamp"] == "2026-04-16T01:00:00Z"
+    assert artifacts[1].metadata["response_timestamp"] == "2026-04-16T01:03:00Z"

tests/test_mcdonald_wizard.py

@@ -0,0 +1,387 @@
+#!/usr/bin/env python3
+"""
+McDonald Wizard Test Suite
+
+Tests for the McDonald chatbot wizard harness and Hermes shim.
+
+Usage:
+    pytest tests/test_mcdonald_wizard.py -v
+    RUN_LIVE_TESTS=1 pytest tests/test_mcdonald_wizard.py -v  # real API calls
+"""
+
+import os
+import sys
+import time
+from pathlib import Path
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+sys.path.insert(0, str(Path(__file__).parent.parent))
+
+from nexus.mcdonald_wizard import (
+    DEFAULT_ENDPOINT,
+    DEFAULT_RETRIES,
+    DEFAULT_TIMEOUT,
+    WIZARD_ID,
+    McdonaldWizard,
+    WizardResponse,
+    mcdonald_wizard,
+)
+
+
+# ═══════════════════════════════════════════════════════════════════════════
+# FIXTURES
+# ═══════════════════════════════════════════════════════════════════════════
+
+
+@pytest.fixture
+def wizard():
+    """Wizard with a fake API key so no real calls are made."""
+    return McdonaldWizard(api_key="fake-key-for-testing")
+
+
+@pytest.fixture
+def mock_ok_response():
+    """Mock requests.post returning a successful API response."""
+    mock = MagicMock()
+    mock.status_code = 200
+    mock.json.return_value = {
+        "choices": [{"message": {"content": "Behold, the golden arches!"}}],
+        "model": "mc-wizard-v1",
+    }
+    return mock
+
+
+@pytest.fixture
+def mock_rate_limit_response():
+    """Mock requests.post returning a 429 rate-limit error."""
+    mock = MagicMock()
+    mock.status_code = 429
+    mock.text = "Rate limit exceeded"
+    return mock
+
+
+@pytest.fixture
+def mock_server_error_response():
+    """Mock requests.post returning a 500 server error."""
+    mock = MagicMock()
+    mock.status_code = 500
+    mock.text = "Internal server error"
+    return mock
+
+
+# ═══════════════════════════════════════════════════════════════════════════
+# WizardResponse dataclass
+# ═══════════════════════════════════════════════════════════════════════════
+
+
+class TestWizardResponse:
+    def test_default_creation(self):
+        resp = WizardResponse()
+        assert resp.text == ""
+        assert resp.model == ""
+        assert resp.latency_ms == 0.0
+        assert resp.attempt == 1
+        assert resp.error is None
+        assert resp.timestamp
+
+    def test_to_dict_includes_all_fields(self):
+        resp = WizardResponse(text="Hello", model="mc-wizard-v1", latency_ms=42.5, attempt=2)
+        d = resp.to_dict()
+        assert d["text"] == "Hello"
+        assert d["model"] == "mc-wizard-v1"
+        assert d["latency_ms"] == 42.5
+        assert d["attempt"] == 2
+        assert d["error"] is None
+        assert "timestamp" in d
+
+    def test_error_response(self):
+        resp = WizardResponse(error="HTTP 429: Rate limit")
+        assert resp.error == "HTTP 429: Rate limit"
+        assert resp.text == ""
+
+
+# ═══════════════════════════════════════════════════════════════════════════
+# McdonaldWizard — initialization
+# ═══════════════════════════════════════════════════════════════════════════
+
+
+class TestMcdonaldWizardInit:
+    def test_default_endpoint(self, wizard):
+        assert wizard.endpoint == DEFAULT_ENDPOINT
+
+    def test_custom_endpoint(self):
+        w = McdonaldWizard(api_key="k", endpoint="https://custom.example.com/chat")
+        assert w.endpoint == "https://custom.example.com/chat"
+
+    def test_default_timeout(self, wizard):
+        assert wizard.timeout == DEFAULT_TIMEOUT
+
+    def test_default_retries(self, wizard):
+        assert wizard.max_retries == DEFAULT_RETRIES
+
+    def test_no_api_key_warning(self, caplog):
+        import logging
+
+        with caplog.at_level(logging.WARNING, logger="mcdonald_wizard"):
+            McdonaldWizard(api_key="")
+        assert "MCDONALDS_API_KEY" in caplog.text
+
+    def test_api_key_from_env(self, monkeypatch):
+        monkeypatch.setenv("MCDONALDS_API_KEY", "env-key-123")
+        w = McdonaldWizard()
+        assert w.api_key == "env-key-123"
+
+    def test_endpoint_from_env(self, monkeypatch):
+        monkeypatch.setenv("MCDONALDS_ENDPOINT", "https://env.example.com/chat")
+        w = McdonaldWizard(api_key="k")
+        assert w.endpoint == "https://env.example.com/chat"
+
+    def test_initial_stats_zero(self, wizard):
+        assert wizard.request_count == 0
+        assert wizard.total_latency_ms == 0.0
+
+
+# ═══════════════════════════════════════════════════════════════════════════
+# McdonaldWizard — ask (mocked HTTP)
+# ═══════════════════════════════════════════════════════════════════════════
+
+
+class TestAsk:
+    def test_ask_no_api_key_returns_error(self):
+        w = McdonaldWizard(api_key="")
+        resp = w.ask("Hello wizard")
+        assert resp.error is not None
+        assert "MCDONALDS_API_KEY" in resp.error
+
+    def test_ask_success(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response):
+            resp = wizard.ask("What is your wisdom?")
+
+        assert resp.error is None
+        assert resp.text == "Behold, the golden arches!"
+        assert resp.model == "mc-wizard-v1"
+        assert resp.latency_ms >= 0.0
+        assert resp.attempt == 1
+
+    def test_ask_increments_request_count(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response):
+            wizard.ask("q1")
+            wizard.ask("q2")
+
+        assert wizard.request_count == 2
+
+    def test_ask_with_system_prompt(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response) as mock_post:
+            wizard.ask("Hello", system="You are a wise McDonald wizard")
+
+        payload = mock_post.call_args[1]["json"]
+        roles = [m["role"] for m in payload["messages"]]
+        assert "system" in roles
+        assert payload["messages"][0]["content"] == "You are a wise McDonald wizard"
+
+    def test_ask_with_context(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response) as mock_post:
+            wizard.ask("Continue please", context="Prior context here")
+
+        payload = mock_post.call_args[1]["json"]
+        contents = [m["content"] for m in payload["messages"]]
+        assert "Prior context here" in contents
+
+    def test_ask_without_optional_args(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response) as mock_post:
+            wizard.ask("Simple prompt")
+
+        payload = mock_post.call_args[1]["json"]
+        assert payload["messages"][-1]["role"] == "user"
+        assert payload["messages"][-1]["content"] == "Simple prompt"
+
+    def test_ask_sends_bearer_auth(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response) as mock_post:
+            wizard.ask("Hello")
+
+        headers = mock_post.call_args[1]["headers"]
+        assert headers["Authorization"] == "Bearer fake-key-for-testing"
+
+    def test_ask_api_failure_returns_error(self, wizard):
+        with patch("requests.post", side_effect=Exception("Connection refused")):
+            resp = wizard.ask("Hello")
+
+        assert resp.error is not None
+        assert "failed" in resp.error.lower()
+        assert wizard.request_count == 1
+
+
+# ═══════════════════════════════════════════════════════════════════════════
+# McdonaldWizard — retry behaviour
+# ═══════════════════════════════════════════════════════════════════════════
+
+
+class TestRetry:
+    def test_retries_on_429(self, wizard, mock_ok_response, mock_rate_limit_response):
+        call_count = [0]
+
+        def side_effect(*args, **kwargs):
+            call_count[0] += 1
+            if call_count[0] < 2:
+                return mock_rate_limit_response
+            return mock_ok_response
+
+        with patch("requests.post", side_effect=side_effect):
+            with patch("time.sleep"):  # suppress actual sleep
+                resp = wizard.ask("Hello")
+
+        assert resp.error is None
+        assert resp.attempt == 2
+        assert call_count[0] == 2
+
+    def test_retries_on_500(self, wizard, mock_ok_response, mock_server_error_response):
+        call_count = [0]
+
+        def side_effect(*args, **kwargs):
+            call_count[0] += 1
+            if call_count[0] < 3:
+                return mock_server_error_response
+            return mock_ok_response
+
+        with patch("requests.post", side_effect=side_effect):
+            with patch("time.sleep"):
+                resp = wizard.ask("Hello")
+
+        assert resp.error is None
+        assert call_count[0] == 3
+
+    def test_all_retries_exhausted_returns_error(self, wizard, mock_rate_limit_response):
+        with patch("requests.post", return_value=mock_rate_limit_response):
+            with patch("time.sleep"):
+                resp = wizard.ask("Hello")
+
+        assert resp.error is not None
+        assert wizard.request_count == 1
+
+    def test_no_retry_on_success(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response) as mock_post:
+            resp = wizard.ask("Hello")
+
+        assert mock_post.call_count == 1
+        assert resp.attempt == 1
+
+
+# ═══════════════════════════════════════════════════════════════════════════
+# McdonaldWizard — session stats
+# ═══════════════════════════════════════════════════════════════════════════
+
+
+class TestSessionStats:
+    def test_initial_stats(self, wizard):
+        stats = wizard.session_stats()
+        assert stats["wizard_id"] == WIZARD_ID
+        assert stats["request_count"] == 0
+        assert stats["total_latency_ms"] == 0.0
+        assert stats["avg_latency_ms"] == 0.0
+
+    def test_stats_after_calls(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response):
+            wizard.ask("a")
+            wizard.ask("b")
+
+        stats = wizard.session_stats()
+        assert stats["request_count"] == 2
+        assert stats["total_latency_ms"] >= 0.0
+        assert stats["avg_latency_ms"] >= 0.0
+
+    def test_avg_latency_calculation(self, wizard, mock_ok_response):
+        with patch("requests.post", return_value=mock_ok_response):
+            wizard.ask("x")
+
+        stats = wizard.session_stats()
+        assert stats["avg_latency_ms"] == stats["total_latency_ms"]
+
+
+# ═══════════════════════════════════════════════════════════════════════════
+# Hermes tool function
+# ═══════════════════════════════════════════════════════════════════════════
+
+
+class TestHermesTool:
+    def test_mcdonald_wizard_tool_returns_dict(self, monkeypatch):
+        mock_resp = WizardResponse(text="I am the wizard", model="mc-v1")
+        mock_wizard = MagicMock()
+        mock_wizard.ask.return_value = mock_resp
+
+        import nexus.mcdonald_wizard as _mod
+
+        monkeypatch.setattr(_mod, "_wizard_instance", mock_wizard)
+
+        result = mcdonald_wizard("What do you know?")
+
+        assert isinstance(result, dict)
+        assert result["text"] == "I am the wizard"
+        assert result["model"] == "mc-v1"
+        assert result["error"] is None
+
+    def test_mcdonald_wizard_tool_passes_system(self, monkeypatch):
+        mock_resp = WizardResponse(text="Aye", model="mc-v1")
+        mock_wizard = MagicMock()
+        mock_wizard.ask.return_value = mock_resp
+
+        import nexus.mcdonald_wizard as _mod
+
+        monkeypatch.setattr(_mod, "_wizard_instance", mock_wizard)
+        mcdonald_wizard("Hello", system="Be brief")
+
+        mock_wizard.ask.assert_called_once_with("Hello", system="Be brief")
+
+    def test_mcdonald_wizard_tool_propagates_error(self, monkeypatch):
+        mock_resp = WizardResponse(error="API key missing")
+        mock_wizard = MagicMock()
+        mock_wizard.ask.return_value = mock_resp
+
+        import nexus.mcdonald_wizard as _mod
+
+        monkeypatch.setattr(_mod, "_wizard_instance", mock_wizard)
+
+        result = mcdonald_wizard("Hello")
+        assert result["error"] == "API key missing"
+
+
+# ═══════════════════════════════════════════════════════════════════════════
+# Live API tests (skipped unless RUN_LIVE_TESTS=1 and MCDONALDS_API_KEY set)
+# ═══════════════════════════════════════════════════════════════════════════
+
+
+def _live_tests_enabled():
+    return (
+        os.environ.get("RUN_LIVE_TESTS") == "1"
+        and bool(os.environ.get("MCDONALDS_API_KEY"))
+    )
+
+
+@pytest.mark.skipif(
+    not _live_tests_enabled(),
+    reason="Live tests require RUN_LIVE_TESTS=1 and MCDONALDS_API_KEY",
+)
+@pytest.mark.integration
+class TestLiveAPI:
+    """Integration tests that hit the real McDonald chatbot API."""
+
+    @pytest.fixture
+    def live_wizard(self):
+        return McdonaldWizard()
+
+    def test_live_ask(self, live_wizard):
+        resp = live_wizard.ask("Say 'McReady' and nothing else.")
+        assert resp.error is None
+        assert resp.text.strip()
+        assert resp.latency_ms > 0
+
+    def test_live_session_stats_update(self, live_wizard):
+        live_wizard.ask("Ping")
+        stats = live_wizard.session_stats()
+        assert stats["request_count"] == 1
+        assert stats["total_latency_ms"] > 0
+
+
+if __name__ == "__main__":
+    pytest.main([__file__, "-v"])

tests/test_spatial_chat_audio.js

@@ -0,0 +1,125 @@
+/**
+ * Tests for SpatialChatAudio component (issue #1544)
+ */
+
+import { SpatialChatAudio } from '../nexus/components/spatial-chat-audio.js';
+
+// Mock DOM and THREE for Node.js testing
+if (typeof document === 'undefined') {
+  global.document = {
+    addEventListener: () => {},
+    removeEventListener: () => {},
+  };
+}
+
+if (typeof THREE === 'undefined') {
+  global.THREE = {
+    Vector3: class {
+      constructor(x=0, y=0, z=0) { this.x = x; this.y = y; this.z = z; }
+      normalize() { return this; }
+      dot() { return 0; }
+      cross() { return this; }
+    },
+    MathUtils: { clamp: (v, min, max) => Math.max(min, Math.min(max, v)) },
+    AudioListener: class {
+      constructor() {
+        this.context = {
+          state: 'running',
+          currentTime: 0,
+          createOscillator: () => ({
+            type: 'sine',
+            frequency: { value: 440 },
+            connect: () => {},
+            start: () => {},
+            stop: () => {},
+            disconnect: () => {},
+            onended: null,
+          }),
+          createGain: () => ({
+            gain: { value: 1, setValueAtTime: () => {}, linearRampToValueAtTime: () => {}, exponentialRampToValueAtTime: () => {}, setTargetAtTime: () => {} },
+            connect: () => {},
+            disconnect: () => {},
+          }),
+          createStereoPanner: () => ({
+            pan: { value: 0, setValueAtTime: () => {}, setTargetAtTime: () => {} },
+            connect: () => {},
+            disconnect: () => {},
+          }),
+          createPanner: () => ({
+            panningModel: '',
+            distanceModel: '',
+            refDistance: 0,
+            maxDistance: 0,
+            rolloffFactor: 0,
+            positionX: { value: 0, setValueAtTime: () => {} },
+            positionY: { value: 0, setValueAtTime: () => {} },
+            positionZ: { value: 0, setValueAtTime: () => {} },
+            connect: () => {},
+            disconnect: () => {},
+          }),
+          resume: () => Promise.resolve(),
+          destination: {},
+        };
+      }
+    },
+  };
+}
+
+function assert(condition, message) {
+  if (!condition) {
+    console.error(`❌ FAILED: ${message}`);
+    process.exit(1);
+  }
+  console.log(`✔ PASSED: ${message}`);
+}
+
+console.log('--- Running SpatialChatAudio Tests ---');
+
+// Test 1: Module exports
+assert(typeof SpatialChatAudio === 'object', 'SpatialChatAudio exports an object');
+assert(typeof SpatialChatAudio.init === 'function', 'SpatialChatAudio has init method');
+assert(typeof SpatialChatAudio.playChatSound === 'function', 'SpatialChatAudio has playChatSound method');
+assert(typeof SpatialChatAudio.createVoiceSource === 'function', 'SpatialChatAudio has createVoiceSource method');
+
+// Test 2: Config defaults
+assert(SpatialChatAudio.isEnabled() === true, 'Enabled by default');
+assert(SpatialChatAudio.getMaxHearingDistance() === 40, 'Default max hearing distance is 40');
+
+// Test 3: Configuration changes
+SpatialChatAudio.setMaxHearingDistance(60);
+assert(SpatialChatAudio.getMaxHearingDistance() === 60, 'Max hearing distance updated to 60');
+
+SpatialChatAudio.setEnabled(false);
+assert(SpatialChatAudio.isEnabled() === false, 'Can disable audio');
+
+SpatialChatAudio.setEnabled(true);
+assert(SpatialChatAudio.isEnabled() === true, 'Can re-enable audio');
+
+// Test 4: Initialization with mock camera
+const mockCamera = {
+  position: new THREE.Vector3(0, 0, 0),
+  getWorldDirection: () => new THREE.Vector3(1, 0, 0),
+  up: new THREE.Vector3(0, 1, 0),
+  add: () => {},
+  getObjectByProperty: () => null,
+};
+
+SpatialChatAudio.init(mockCamera);
+assert(true, 'SpatialChatAudio initializes with camera');
+
+// Test 5: Voice source creation
+const voiceSource = SpatialChatAudio.createVoiceSource('timmy', new THREE.Vector3(10, 0, 0));
+assert(voiceSource !== null, 'Voice source created');
+assert(voiceSource.agentId === 'timmy', 'Voice source has correct agentId');
+assert(typeof voiceSource.updatePosition === 'function', 'Voice source has updatePosition');
+assert(typeof voiceSource.destroy === 'function', 'Voice source has destroy');
+
+// Test 6: Voice source position update
+voiceSource.updatePosition(new THREE.Vector3(20, 0, 0));
+assert(true, 'Voice source position updated');
+
+// Test 7: Voice source cleanup
+voiceSource.destroy();
+assert(true, 'Voice source destroyed');
+
+console.log('--- All SpatialChatAudio Tests Passed ---');