v3.7.0

3.7.0 (2026-07-02)

Bug Fixes

• ci: update Sentinel to warn on clear-named skills, remove bridge directories (e76a7d5)
• skills: add bridge docs/members/ dirs for clear-named skills (ddc45d0)

Features

• skills: add clear-named skill mirrors and shared reference checklists (a1da6ad), closes #346

v3.6.0

3.6.0 (2026-07-01)

Bug Fixes

• core: address Warden findings — extract memberLore, refactor run.ts, deduplicate constants, add injection guards (1fda655)
• core: address Warden findings round 3 — indentation, dead code, boolean naming, nesting, unused dep (00b5279)
• core: address Warden/Auditor/Reviewer findings round 2 (b5923ed)

Features

• core: rename src/skills to src/tools, add skill discovery infrastructure (e4eed6b)

v3.5.3

3.5.3 (2026-07-01)

Bug Fixes

• groq: update default model from decommissioned llama-3.1-70b to llama-3.3-70b-versatile (483c54a)
• groq: update test expectation to match default model llama-3.3-70b-versatile (0593405)

v3.5.2

3.5.2 (2026-06-29)

Bug Fixes

• ollama: read baseUrl from config before falling back to env (75ecd2b)

v3.5.1

3.5.1 (2026-06-29)

Bug Fixes

• align 3 remaining stale doc references (4213b4c)
• update all project-wide references after moving doc dirs to docs/ (83cc7c5)

v3.5.0

3.5.0 (2026-06-28)

Bug Fixes

• ci: add explicit return 0 in decision function and guard call (1eeabc6)
• ci: add explicit return 0 in else branch for consistency (bb26af5)
• ci: add explicit return 0 to check_agenthood_decision (bd3f78a)
• ci: address auditor findings — test-runner flag injection, npm ci, npm audit, temp paths (67e9f3e)
• ci: address auditor high and medium findings (b2bb7ac)
• ci: address review findings — setup-node in action, smart tester, extract scripts, pin tools (be12b18)
• ci: address reviewer and warden findings (ac10c87)
• ci: address warden stale-comment quoting, temp_dir casing, revert vscode npm ci (d7ec7b2)
• ci: align blocking detection in reviewer.yml with line-start grep pattern (307c75e)
• ci: anchor blocking regex, ignore-scripts, extract helpers (0969637)
• ci: anchor blocking=true grep to require closing --> (aae8c1f)
• ci: change AGENTHOOD_DECISION format to avoid false-positive true|false (672e1dd)
• ci: export OPENCODE_API_KEY, add setup-node cache, smart test selection (d03f924)
• ci: fail workflows on blocking findings, restore api-key, fix registry paths, add execute permission (b60e63e)
• ci: install commitlint packages locally for config resolution (0d3046f)
• ci: match [blocking] in summary table cells not line start for reviewer (65851d8)
• ci: narrow stale-comment matching, remove awk guard (7152e36)
• ci: prevent set -e from killing script on CLI failure (700f36d)
• ci: resolve commitlint tsx resolution and librarian false-positive (5a2bd81)
• ci: resolve TASK unbound variable from subshell scoping and commitlint tsx resolution (07df150)
• ci: restrict blocking check to table rows only (9b562f7)
• ci: revert stale_previous_comment --arg in jq filter, add pipefail guard (1495f4a)
• ci: tighten blocking detection to line-start grep, bump header-max-length to 150 (44d9db0)
• ci: use AGENTHOOD_DECISION for stale comment matching (563665c)

Features

• ci: mark previous agent PR comments as outdated instead of deleting (2c3b6fa)
• ci: use structured decision block from LLM for blocking detection (ab32887)

v3.4.0

3.4.0 (2026-06-28)

Bug Fixes

• evals: import ExecutionContext from correct module (4176542)
• evals: import ExecutionContext from correct module in tests (eeb9333)

Features

• evals: implement EpisodeLearner — update LongTermMemory and ResidualMemory from eval scores (b881c53), closes #119

v3.3.0

3.3.0 (2026-06-28)

Bug Fixes

• correct vitest JSON field names (num prefix), remove duplicate detect block in run.ts (d98c3eb)

Features

• core: implement metrics collector, status --watch/--json/--drift, quality gates drift detection and config (32cd8ee)

v3.2.0

3.2.0 (2026-06-28)

Bug Fixes

• address all review findings across 7 phase branches (926b8c8)
• remove hardcoded member count from steward readme (maintenance trap) (4c25127)
• steward pre-load count should be 15 (16 members - 1 for steward itself) (a48435d)
• sync skills/ directory with members/ for ci compliance (2e4a655)
• use 16 not 15 for steward pre-load count (matches registry) (efa5e9b)

Features

• phase: phase 0 - decision log, postmortem, auto-discover (950ed85), closes #279 #280 #114
• phase: phase 1 - protocol interfaces and workflow engine (5c700e8), closes #116 #116
• phase: phase 2 - workflow checkpoint and goal chain (f3160ba), closes #117 #118
• phase: phase 3 - oracle, strategist, and operator agents (a886be1), closes #113 #277 #278
• phase: phase 4 - verify, rollback, and status commands (ce1c560), closes #275 #276 #281
• phase: phase 5 - diff impact analyzer and quality gates (26fc8a1), closes #115 #282
• phase: phase 6 - review-pr workflow end-to-end (ea0debf)

v3.1.0

3.1.0 (2026-06-27)

Bug Fixes

• docs: address Reviewer findings on init check count and LanceDBStore API example (4c215ef), closes #286
• docs: correct check count to 21 and fix insert->add API example (deb2591), closes #286
• docs: fix mentioned shipped version (60a43e1)
• init,check: align init ceremony with health check expectations (cd0a6bc), closes #286
• skills: add output format section to the-reviewer SKILL.md for consistent rendering (86c7162), closes #286
• sync skills/the-reviewer/SKILL.md with members/ changes (473707d), closes #286
• the-reviewer: address review findings on output format and README (8aa91f9), closes #286
• the-reviewer: flatten heading hierarchy and add intra-section spacing example (faa73fc), closes #286
• the-reviewer: use [SEVERITY] placeholder and move meta-instruction outside template (e073edb), closes #286

Features

• load .env file automatically via dotenv (91fac79), closes #286