docs: QA triage action plan for #691

Structured action plan converting cross-repo QA findings into
executable steps with owners, priorities, and verification.

Key findings addressed:
- P0: Production surfaces down (DNS/nginx), playground broken
- P1: 166 open PRs across 5 repos, 58 issues with duplicates
- P2: the-door crisis features blocked, no branch protection
- P3: Burn dedup gate, nightly triage cron

Priority order:
1. Fix DNS/nginx (crisis intervention reachable)
2. Close duplicate PRs (clear noise)
3. Review the-door PRs (mission-critical)
4. Fix the-playground (user-facing)
5. Enable branch protection
6. Build dedup gate
7. Nightly triage cron

Closes #691.

reports/qa-triage/2026-04-14-action-plan.md

@@ -0,0 +1,218 @@
+# QA Triage Action Plan — Foundation-Wide (2026-04-14)
+
+> **Source:** Issue #691 — Cross-Repo Deep QA Report
+> **Generated:** 2026-04-14
+> **Status:** Active triage — actionable steps for each finding
+
+---
+
+## Executive Summary
+
+The QA sweep identified systemic issues across the Foundation. Current state (verified live):
+
+| Metric | QA Report | Current | Trend |
+|--------|-----------|---------|-------|
+| Total open PRs | ~55+ | **166** | Worsening |
+| Repos with dupes | 3 | **5 (all)** | Worsening |
+| Duplicate PR issues | 7+ | **58** | Critical |
+| Prod surfaces reachable | 0/4 | 0/4 | Unchanged |
+
+**The core problem:** Burn sessions generate faster than triage can absorb. The backlog is growing, not shrinking.
+
+---
+
+## P0 — Critical
+
+### 1. Production Surfaces Down (404 on all endpoints)
+
+**Status:** Unchanged since QA report
+**Impact:** Zero users can reach any Timmy surface. The Door (crisis intervention) is unreachable.
+
+| Surface | URL | Status |
+|---------|-----|--------|
+| Root | http://143.198.27.163/ | nginx 404 |
+| Nexus | http://143.198.27.163/nexus/ | 404 |
+| Playground | http://143.198.27.163/playground/ | 404 |
+| Tower | http://143.198.27.163/tower/ | 404 |
+| Domain | https://alexanderwhitestone.com/ | DNS broken |
+
+**Action:**
+- [ ] Verify DNS records for alexanderwhitestone.com (check registrar)
+- [ ] SSH to VPS, check nginx config: `nginx -T`
+- [ ] Ensure server blocks exist for each location
+- [ ] Restart nginx: `systemctl restart nginx`
+- [ ] Tracked in the-nexus#1105
+
+**Owner:** Infrastructure
+**Priority:** Immediate — this is the mission
+
+### 2. the-playground index.html Broken
+
+**Status:** Unconfirmed since QA report
+**Impact:** Playground app crashes on load — missing script tags
+
+**Action:**
+- [ ] Read the-playground/index.html
+- [ ] Verify script tags for all JS modules
+- [ ] Fix missing imports
+- [ ] Tracked in the-playground#200
+
+**Owner:** the-playground
+**Priority:** High — blocks user-facing playground
+
+---
+
+## P1 — High (Duplicate PR Crisis)
+
+### 3. Duplicate PR Storm Across All Repos
+
+**Current state (verified live 2026-04-14):**
+
+| Repo | Open PRs | Issues with Duplicates | Worst Case |
+|------|----------|----------------------|------------|
+| the-nexus | 44 | 16 | Issue #1509 → 4 PRs |
+| the-playground | 31 | 10 | Issue #180 → 3 PRs |
+| the-door | 27 | 6 | Issue #988 → 7 PRs |
+| timmy-config | 50 | 20 | Issue #50 → 7 PRs |
+| timmy-home | 14 | 6 | Issue #50 → 6 PRs |
+| **Total** | **166** | **58 issues** | — |
+
+**Root cause:** Burn sessions create branches without checking for existing PRs on the same issue. No deduplication gate in the burn pipeline.
+
+**Immediate action — close duplicates per repo:**
+
+For each issue with multiple PRs:
+1. Keep the PR with the most commits/diff (most complete implementation)
+2. Close all others with comment: "Closing duplicate. See #PR for primary implementation."
+3. If no PR is clearly superior, keep the oldest (first mover)
+
+**Script to identify duplicates:**
+```bash
+# For each repo, list issues with >1 open PR
+python3 scripts/duplicate-pr-detector.py --repo <repo> --close-duplicates
+```
+
+**Long-term fix:**
+- [ ] Add pre-flight check to burn loop: query open PRs before creating new branch
+- [ ] Add Gitea label `burn-active` to track which issues have active burn PRs
+- [ ] Add CI check that rejects PR if another open PR references the same issue
+
+**Owner:** Fleet / Burn infrastructure
+**Priority:** High — duplicates waste review time and create merge conflicts
+
+### 4. Misfiled PR in wrong repo
+
+**the-nexus PR #1521:** "timmy-home Backlog Triage Report" is filed in the-nexus but concerns timmy-home.
+
+**Action:**
+- [ ] Close PR #1521 in the-nexus with redirect comment
+- [ ] File content as issue or PR in timmy-home if still relevant
+
+---
+
+## P2 — Medium
+
+### 5. the-door Crisis Features Blocked
+
+Mission-critical PRs sitting unreviewed:
+
+| Issue | Title | Impact |
+|-------|-------|--------|
+| #91 | Safety plan improvements | User safety |
+| #89 | Safety plan enhancements | User safety |
+| #90 | Crisis overlay fixes | UX |
+| #87 | Crisis overlay bugs | UX |
+| 988 link | Crisis hotline link fix | **Life safety** |
+
+**Action:**
+- [ ] Prioritize the-door PR review over all other repos
+- [ ] Assign a reviewer or run dedicated triage session for the-door only
+- [ ] After review, merge in dependency order
+
+**Owner:** Crisis team / Alexander
+**Priority:** High — this is the mission
+
+### 6. Branch Protection Missing Foundation-Wide
+
+No repo has branch protection enabled. Any member can push directly to main.
+
+**Action:**
+- [ ] Enable branch protection on all repos with:
+  - Require 1 approval before merge
+  - Require CI to pass (where CI exists)
+  - Dismiss stale approvals on new commits
+- [ ] Covered in timmy-home PR #606 but not yet implemented
+
+**Repos without CI (need smoke test first):**
+- the-playground
+- the-beacon
+- timmy-home
+
+**Owner:** Alexander / Infrastructure
+**Priority:** Medium — prevents accidental breakage
+
+---
+
+## P3 — Low (Process Improvements)
+
+### 7. Burn Session Deduplication Gate
+
+**Problem:** Burn loops don't check for existing PRs before creating new ones.
+
+**Solution:** Pre-flight check in burn pipeline:
+```python
+def has_open_pr(owner, repo, issue_number):
+    prs = gitea.get_pulls(owner, repo, state="open")
+    for pr in prs:
+        if f"#{issue_number}" in (pr.get("body", "") or ""):
+            return True
+    return False
+```
+
+**Action:**
+- [ ] Add to hermes-agent burn loop
+- [ ] Add to timmy-config burn scripts
+- [ ] Test with dry-run before enabling
+
+### 8. Nightly Triage Cron
+
+**Problem:** No automated triage. Duplicates accumulate until manual sweep.
+
+**Solution:** Nightly cron that:
+1. Scans all repos for duplicate PRs
+2. Posts summary to a triage channel
+3. Auto-closes duplicates older than 48h with lower diff count
+
+**Action:**
+- [ ] Design triage cron job spec
+- [ ] Implement as hermes cron job
+- [ ] Run nightly at 03:00 UTC
+
+---
+
+## Priority Order (Execution Sequence)
+
+1. **Fix DNS/nginx** — The Door must be reachable (crisis intervention = the mission)
+2. **Close duplicate PRs** — 58 issues with dupes, clear the noise
+3. **Review the-door PRs** — Mission-critical crisis features
+4. **Fix the-playground** — User-facing app broken
+5. **Enable branch protection** — Prevent future breakage
+6. **Build dedup gate** — Prevent future duplicate storms
+7. **Nightly triage cron** — Automated hygiene
+
+---
+
+## Verification Checklist
+
+After completing actions above, verify:
+
+- [ ] http://143.198.27.163/ returns a page (not 404)
+- [ ] https://alexanderwhitestone.com/ resolves
+- [ ] All repos have <5 duplicate PRs
+- [ ] the-door has 0 unreviewed safety/crisis PRs
+- [ ] Branch protection enabled on all repos
+- [ ] Burn loop has pre-flight PR check
+
+---
+
+*This plan converts QA findings into executable actions. Each item has an owner, priority, and verification step.*