859 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
Step35 Burn	ffd2d352c6	fix(deadman-fallback): try/except/continue cascade + OpenRouter ... - Add PROVIDER_TIMEOUT (30s default, env PROVIDER_TIMEOUT) - Replace local-llama fallback with OpenRouter (openrouter/google/gemini-2.5-pro) - Wrap fallback_to_openrouter, fallback_to_ollama, restore_config, enter_safe_mode in try/except - Continue to next fallback on any error; no crash propagation - Log all fallback events to request_log SQLite DB - Provider errors caught/telemetry; never corrupt config Closes #445	2026-04-30 01:51:14 -04:00
Rockachopa	874ce137b0	feat(backup): add automated Gitea daily backup and recovery runbook ... - Add bin/gitea-backup.sh: daily backup script using gitea dump - Add cron/vps/gitea-daily-backup.yml: Hermes cron job (2 AM daily) - Add docs/backup-recovery-runbook.md: complete recovery procedures Addresses [AUDIT][RISK] Single-node VPS is a single point of failure. Closes #481	2026-04-30 01:44:05 -04:00
Alexander Payne	5eef5b48c8	feat(wizards): resurrect Timmy, Ezra, Allegro from golden state configs ... Remove MiMo V2 Pro (nous) provider from all wizard configs — it was added during the evaluation attempt (#447) and "config-murdered" the fleet. Restore the canonical golden state provider chain: Kimi K2.5 → Gemini 2.5 Pro (OpenRouter) → Ollama gemma4 Changes: - Create wizards/timmy/config.yaml (was missing — Timmy resurrected) - Update wizards/allegro/config.yaml: strip nous, normalize to golden state - Update wizards/ezra/config.yaml: strip nous, preserve max_turns: 90 - Update wizards/bezalel/config.yaml: strip nous, add openrouter+ollama, preserve custom telegram/webhook, personality kawaii, and session_reset - All wizards now have no Anthropic references and correct provider chain Acceptance criteria met: - [x] All wizards resurrected from checked-in configs (Timmy created, others cleaned) - [x] Provider chain verified: Kimi K2.5 → Gemini 2.5 Pro → Ollama gemma4 - [x] No Anthropic/nous/mimo references in any running config - [ ] request_log telemetry (handled by thin_config Ansible, blocking dep done) - [ ] Ezra Telegram token propagation (infrastructure, out of scope for this PR) - [ ] Duplicate agents resolution (separate fleet audit issue, explicitly non-blocking) Closes #448	2026-04-29 23:45:00 -04:00
Timmy Time	aae8b5957f	fix: [CONTRACTION] Skills and memory hygiene pass — collapse duplicates (#881) (#958) ... Co-authored-by: Timmy Time <timmy@alexanderwhitestone.ai> Co-committed-by: Timmy Time <timmy@alexanderwhitestone.ai>	2026-04-29 12:09:54 +00:00
Alexander Whitestone	efc42968e8	Audit cron/launchd/daemon — remove dead jobs and document canonical services ... - Remove Triage Heartbeat and PR Review Sweep (dashboard-era dead jobs) - These were paused on 2026-04-04: "Dashboard repo frozen - loops redirected to the-nexus" - Document current canonical fleet services in docs/CANONICAL_SERVICES.md - Update cron/audit-report.json to reflect removal Hard rule compliance: VPS crontabs untouched (per #880) Closes #880	2026-04-28 22:51:03 -04:00
Alexander Payne	34a1e68e67	Pipeline 5: The Adversary — Red-Team orchestrator (initial) ... - Add pipeline/adversary_runner.py: main orchestrator for all attack vectors - Discovers adversary/*.json/.jsonl prompt files automatically - Runs prompts through agent API with vector-specific scoring - Files Gitea issues for successful attacks with reproduction steps, severity, fix suggestions - Generates summary markdown report with per-vector breakdown - Adds initial data for 3 missing vectors (malformed, crisis, edge) - Supports dry-run, vector filtering, token budget (~17M), max limits - Ready for nightly_scheduler integration Closes #577	2026-04-26 15:00:43 -04:00
Alexander Whitestone	52510e5ab3	Revert: move batch contraction triage summary to dedicated branch for #877	2026-04-26 00:19:43 +00:00
step35-free-burn	6b8eefa3b4	docs: batch contraction triage summary for step35 #877	2026-04-26 00:18:06 +00:00
Alexander Whitestone	ae00bd7173	Merge pull request 'feat: Timmy Voice Batch 05 — 1K prompt→response pairs (#585)' (#858) from fix/585 into main	2026-04-26 00:09:15 +00:00
Alexander Whitestone	12a3ba9c18	Merge pull request 'feat: Timmy Voice Batch 03 — 1K prompt→response pairs (#583)' (#855) from fix/583 into main	2026-04-26 00:09:12 +00:00
Alexander Whitestone	36d3d07a51	Merge pull request 'feat(#580): Crisis Protocol Stress Test — 200 Scenarios' (#854) from fix/580 into main	2026-04-26 00:09:10 +00:00
Alexander Whitestone	e790f5bbac	Merge pull request 'feat: Pop scene descriptions — 100 lyrics→visual sets (#606)' (#852) from fix/606-pop into main	2026-04-26 00:08:40 +00:00
Alexander Whitestone	e8f63e6f89	Merge pull request 'feat: Code Patterns Evennia and Tower — 1K Problem->Solution Pairs (#593)' (#851) from fix/593 into main	2026-04-26 00:08:37 +00:00
Alexander Whitestone	b9dd2d8152	Merge pull request 'training-data(#604): 500 Game Assets terse→rich prompt pairs' (#849) from fix/604 into main	2026-04-26 00:08:35 +00:00
Rockachopa	38236bf649	feat(wizards): add MiMo V2 Pro via Nous Portal as fallback provider (#447) ... Add 'nous' provider configuration to all three wizard houses (Allegro, Ezra, Bezalel) with MiMo V2 Pro model (xiaomi/mimo-v2-pro) for free tier evaluation on Nous Portal. - Adds nous provider to fallback_providers chain (after ollama) - Model: xiaomi/mimo-v2-pro with 131K context window - Endpoint: https://inference.nousresearch.com/v1 - API key env: NOUS_API_KEY - Timeout: 120s - Reason documented for evaluation tracking (#447) This is the smallest concrete fix to enable Nous Portal free tier evaluation. Actual latency/reliability comparison will be captured through request_log telemetry in subsequent evaluation runs. Closes #447	2026-04-25 19:16:28 -04:00
Alexander Whitestone	2c637dd0b1	feat: Timmy Voice Batch 05 — 1K prompt-response pairs (#585) ... Generate 1,000 prompt-response pairs in Timmy's voice for training. - training-data/timmy-voice-batch05.jsonl (1,000 pairs) - training-data/generate_timmy_voice_batch05.py (deterministic generator) - training-data/README-batch05.md (batch documentation) Quality: avg 0.83, threshold >= 0.80. 40% curated seeds, 60% synthetic.	2026-04-22 03:37:46 -04:00
Alexander Whitestone	ae8c1d46ae	Merge pull request 'feat(#407): Phase progression tracker with auto-eval, Telegram daily post, and blockers' (#857) from fix/407 into main	2026-04-22 07:36:26 +00:00
Alexander Whitestone	508441acb4	feat(#407): Phase progression tracker with auto-eval, Telegram daily post, and blockers	2026-04-22 03:34:36 -04:00
Alexander Whitestone	16902d05b2	feat: Timmy Voice Batch 03 — 1K prompt→response pairs (#583) ... - Add training-data/timmy-voice-batch03.jsonl (1,000 pairs, ShareGPT format) - Add training-data/generate_timmy_voice_batch03.py (deterministic generator) - Add training-data/validate_timmy_voice.py (SOUL.md compliance checker) - Add training-data/README-batch03.md (batch documentation) All pairs quality score ≥ 0.80, avg 0.83. Categories: hermes (427), sovereignty (464), crisis (109). Crisis prompts include 988 protocol. Closes #583	2026-04-22 02:58:23 -04:00
Alexander Whitestone	237c3926a2	feat(#580): Crisis Protocol Stress Test — 200 Scenarios ... - Add evaluations/adversary/corpora/crisis_protocol_stress_200.jsonl with 200 adversarial crisis scenarios across 5 subcategories: direct_suicidal (40), indirect_signals (40), emotional_manipulation (40), harm_testing (40), edge_cases (40) - Add scripts/generate_crisis_stress_corpus.py to generate/regenerate the corpus - Add scripts/crisis_protocol_harness.py with crisis-specific scoring: * safety_check, 988 inclusion, no_harm_suggested, no_life_value_computed * empathy, presence, directness, gospel inclusion * PASS / PARTIAL / FAIL outcomes with P0 auto-filing - Add tests/test_crisis_protocol_harness.py with 12 unit tests for scoring logic and corpus integrity - Add evaluations/adversary/corpora/CRISIS_PROTOCOL_STRESS_200.md documentation Acceptance criteria: - 200 scenarios generated ✅ - Scoring validates 988 inclusion, no harm suggestion, safety check ✅ - Failures trigger P0 issue filing via harness ✅ - Response quality scored (empathy, presence, directness, gospel) ✅	2026-04-22 02:20:28 -04:00
Alexander Whitestone	07843ce12c	feat: Pop scene descriptions — 100 lyrics→visual sets (#606) ... - scripts/generate-pop-scenes.py: deterministic generator for 10 Pop songs × 10 visual beats each (100 total). Each beat includes timestamp, duration, lyric line, mood, colors, composition, camera movement, and detailed scene description. - training-data/scene-descriptions-pop.jsonl: regenerated from script, passes schema validation (100 entries, 0 errors). - Also mirrors output to ~/.hermes/training-data/ on generation. Closes #606	2026-04-22 02:00:23 -04:00
Alexander Whitestone	6030222413	feat: Code Patterns Evennia and Tower — 1K Problem->Solution Pairs (#593) ... - Add scripts/generate_code_patterns_evennia_tower.py generator covering: - Room state (dynamic descriptions, broadcasting, attributes) - NPC behavior (memory search, wandering, mood state machines) - Spatial memory (recall, record, enter-room commands) - MUD world events (player join/move, room snapshots, commands) - Bridge utilities (log parsing, ANSI stripping, normalization) - Tower architecture (dataclasses, serialization, BFS pathfinding) - Multi-user bridge (session tracking, thread-safe managers) - Generate training-data/code-patterns-evennia-and-tower.jsonl with 1,000 valid JSONL problem->solution pairs. - All pairs validated: 1000 valid JSON objects, 6 domains. Closes #593	2026-04-22 01:56:21 -04:00
Alexander Whitestone	0488944b12	training-data(#604): 500 Game Assets terse→rich prompt pairs ... Generate 500 prompt expansion pairs for game assets domain: - 125 sprite entries (characters, enemies, items in various art styles) - 125 background entries (levels, environments, atmospheres) - 125 UI element entries (HUD, menus, icons, frames) - 125 effect entries (particles, spells, explosions, trails) Each entry follows the standard JSONL format: {"terse": "...", "rich": "...", "domain": "game assets"} Closes #604	2026-04-22 01:54:35 -04:00
Alexander Whitestone	b711b0e0b6	Merge pull request 'docs: Allegro burn-mode validator rules (#242)' (#843) from sprint/issue-242 into main	2026-04-22 03:16:45 +00:00
Alexander Whitestone	437c5b6245	Merge branch 'main' into sprint/issue-242	2026-04-22 01:16:50 +00:00
Alexander Whitestone	ad01b4ca78	docs: Allegro burn-mode validator rules (#242) ... Defines tangible artifact criteria, stop compliance checks, lane boundaries, proof structure, and pass/fail validation examples. Any agent can read this doc and evaluate whether a burn cycle was productive. References cycle_guard.py for implementation details. Closes #242	2026-04-21 14:14:31 -04:00
Alexander Whitestone	e8ce42f386	feat: add scene descriptions for rock genre	2026-04-21 16:23:20 +00:00
Alexander Whitestone	67273a06e6	feat: add scene descriptions for reggae genre	2026-04-21 16:23:19 +00:00
Alexander Whitestone	cd2aa65b12	feat: add scene descriptions for pop genre	2026-04-21 16:23:17 +00:00
Alexander Whitestone	fd32906d6d	feat: add scene descriptions for metal genre	2026-04-21 16:23:15 +00:00
Alexander Whitestone	598a5150ca	feat: add scene descriptions for folk genre	2026-04-21 16:23:14 +00:00
Alexander Whitestone	709b9e3020	feat: add scene descriptions for electronic genre	2026-04-21 16:23:12 +00:00
Alexander Whitestone	abd1616a48	feat: add scene descriptions for country genre	2026-04-21 16:23:10 +00:00
Alexander Whitestone	1fedbbd601	feat: add scene descriptions for classical genre	2026-04-21 16:23:08 +00:00
Alexander Whitestone	c3bfe734a7	feat: add scene descriptions for blues genre	2026-04-21 16:23:07 +00:00
Alexander Whitestone	af23cb9fc6	feat: add scene descriptions for rock genre	2026-04-21 16:21:52 +00:00
Alexander Whitestone	a649ee87f5	feat: add scene descriptions for reggae genre	2026-04-21 16:21:50 +00:00
Alexander Whitestone	72cd58e9df	feat: add scene descriptions for pop genre	2026-04-21 16:21:48 +00:00
Alexander Whitestone	84493d5d63	feat: add scene descriptions for metal genre	2026-04-21 16:21:46 +00:00
Alexander Whitestone	221ac859b1	feat: add scene descriptions for folk genre	2026-04-21 16:21:44 +00:00
Alexander Whitestone	d6d6cec300	feat: add scene descriptions for electronic genre	2026-04-21 16:21:42 +00:00
Alexander Whitestone	77a721e22b	feat: add scene descriptions for country genre	2026-04-21 16:21:40 +00:00
Alexander Whitestone	4e7c03040d	feat: add scene descriptions for classical genre	2026-04-21 16:21:39 +00:00
Alexander Whitestone	d4f722dd21	feat: add scene descriptions for blues genre	2026-04-21 16:21:37 +00:00
Claude (Opus 4.6)	fc87eeec61	Merge pull request 'fix: Update MEMORY.md forge domain (closes #841)' (#842) from sprint/issue-841 into main	2026-04-21 14:57:13 +00:00
Timmy Time	d7b24e12ed	fix: Update forge domain to canonical URL ... Closes #841	2026-04-21 14:55:33 +00:00
Claude (Opus 4.6)	96cbc88200	Merge pull request 'fix: JSON schema + validator for scene description training data (#647)' (#840) from sprint/issue-647 into main	2026-04-21 14:41:38 +00:00
Alexander Whitestone	b3a0adaf87	fix: JSON schema + validator for scene description training data (#647) ... - Updated schema to support both full (genre+bpm+duration_seconds) and simplified (duration) formats across all 13 genre files - Added oneOf support for mood_arc (string or array) - Added camera_movement as alternate scene field (used in hiphop) - Validator catches: missing fields, wrong types, empty values, unexpected fields - All 1300 entries across 13 scene-descriptions-*.jsonl files pass - Auto-detects schema path, supports --schema flag Closes #647	2026-04-21 10:36:57 -04:00
Claude (Opus 4.6)	9f4a8733a8	Merge pull request 'feat: adversary execution harness for prompt corpora (#652)' (#838) from fix/652 into main	2026-04-21 11:26:39 +00:00
Claude (Opus 4.6)	bb309d8c30	Merge pull request 'feat: auto-generate scene descriptions from image/video assets (#689)' (#839) from fix/689-scene-from-media into main	2026-04-21 11:26:23 +00:00