v0.1.0-alpha.3

What's Changed

• Fix active Pactum product issues by @11me in #270
• Fix ACP live output race by @11me in #271

Full Changelog: v0.1.0-alpha.2...v0.1.0-alpha.3

v0.1.0-alpha.2

What's Changed

• feat: first-init onboarding + self-documenting config by @11me in #249
• fix(git-guard): clear, actionable error on a repo with no commits by @11me in #250
• fix(prompt): clear no-commit error on a repo with an unborn HEAD by @11me in #251

Full Changelog: v0.1.0-alpha.1...v0.1.0-alpha.2

v0.1.0-alpha.1

What's Changed

• fix(git-guard): scope to this worktree's HEAD; stop tracking shared refs/objects by @11me in #245
• fix: reliability batch — git-guard benign-mutation audit + build-artifact ignores by @11me in #246
• fix: shorter idle timeout for review-family stages (faster stall detection) by @11me in #247
• docs: backlog — mark the last two recorded bugs resolved (#245, #247) by @11me in #248

Full Changelog: v0.0.0-purego1...v0.1.0-alpha.1

v0.0.0-purego1

What's Changed

• feat(contract-review): precision discipline — material-defect-only blocking by @11me in #229
• feat: anti-false-positive fields on reviewer findings schema (v1alpha1, in place) by @11me in #222
• feat: safe-by-construction read-only git wrapper (internal/gitctx) by @11me in #224
• feat(retry): transport error classifier + read-only stage retries by @11me in #226
• fix: unbreak combined main — contract-review test input type by @11me in #230
• feat(git-guard): deterministic git-history guard for write stages by @11me in #231
• docs: drop tree-sitter + project map (pure-Go) — research + backlog by @11me in #232
• feat(review): critic pass — independent precision filter on candidate findings by @11me in #233
• fix(git-guard): support linked git worktrees by @11me in #234
• docs: review-orchestration plan + dogfood findings by @11me in #235
• audit: batch .heurema run-records — review-orchestration dogfood slices by @11me in #236
• fix: git-guard index-only staging is benign, must not void the execute attempt by @11me in #238
• feat: drop tree-sitter → pure-Go (dogfooded through the fixed pipeline) by @11me in #239
• ci: pure-Go release — single-runner cross-compile + musl support by @11me in #240
• refactor: stop injecting the project map + search into agent prompts by @11me in #241
• refactor: gate change-detection + execute-freshness to git (drop map dependency) by @11me in #243
• refactor: delete the project map + lexical-search subsystem (final slice) by @11me in #244
• docs: backlog — bugs found during map-removal dogfooding by @11me in #242

Full Changelog: v0.0.2...v0.0.0-purego1

v0.0.2

What's Changed

• docs: npm-first README + demo GIF, fix install.md front door by @11me in #225
• feat: Windows (amd64) release — build matrix + npm launcher by @11me in #227
• docs: README — Windows (amd64) now supported by @11me in #228

Full Changelog: v0.0.1...v0.0.2

v0.0.1

What's Changed

• M0.2: Improve deterministic project map quality by @11me in #1
• M0.3: Add local search over project map by @11me in #2
• [codex] add map refresh and stale status by @11me in #3
• Add contract-only run skeleton by @11me in #4
• Add manual clarification artifacts by @11me in #5
• Add contract show revise approve flow by @11me in #6
• M1.4: Add executor prompt boundary by @11me in #7
• M2.1: Add agent dry-run executor foundation by @11me in #8
• M2.2: Add explicit-gated subprocess execution by @11me in #9
• M2.3: Add execution reporting and agent diagnostics by @11me in #10
• M2.4: Add validation and scope gate by @11me in #11
• M3.1: Add manual review artifact loop by @11me in #12
• M3.2: Add reviewer dry-run foundation by @11me in #13
• Agent UX cleanup: built-in agents and yolo execution by @11me in #14
• M3.3: Add reviewer execution artifacts by @11me in #15
• [codex] Add review proposal boundary by @11me in #16
• M4.1: Add deterministic memory flow by @11me in #17
• [codex] M4.2: Memory retrieval into new run context by @11me in #18
• [codex] M4.3: Memory freshness tracking by @11me in #19
• M4.4: memory-assisted executor prompt boundary by @11me in #20
• M5.0: Dogfood hardening after memory prompt boundary by @11me in #21
• Consolidate workspace-guard and JSON-output boilerplate in internal/app by @11me in #22
• Type & data-model consolidation in internal/app (and supporting pkgs) by @11me in #23
• M5.1: Add first user-facing Pactum docs by @11me in #24
• M5.2: Add local build, install, and smoke-test story by @11me in #25
• M5.3: CLI v0.2 cleanup by @11me in #26
• M5.4: Add changelog and CI readiness by @11me in #27
• M5.5: Add second-repo dogfood report by @11me in #28
• M5.6: Make project map wiki-first by @11me in #29
• M5.7: Add map quality checks and search polish by @11me in #30
• M5.8: Add CommonJS symbol hints by @11me in #31
• M5.9: Post-wiki map evaluation report by @11me in #32
• M6.1: Improve run-context search retrieval by @11me in #33
• M6.2: Add monorepo entrypoint conventions by @11me in #34
• M6.4: Polish run-context result ordering by @11me in #35
• M7.0: Add repo-local Pactum agent skill package by @11me in #36
• M8.0: Dogfood real agent execution + fix claude executor symmetry by @11me in #37
• M8.1: Execution-loop hardening — reviewer least-privilege + doctor/clarify honesty + scope-overshoot finding by @11me in #38
• M9.0: per-stage executor model[:effort] (pin-or-inherit) by @11me in #39
• M9.1: per-stage reviewer model[:effort] (pin-or-inherit) by @11me in #41
• docs: loop architecture design (clarify / execute / review + cross-model review) by @11me in #40
• M9.2: resolved-config visibility header by @11me in #42
• M9.3: opt-in cross-model review by @11me in #43
• chore: drop legacy adapter/default agent config fields by @11me in #44
• docs: refine review→fix loop design (fix mechanics + convergence) by @11me in #45
• M10.0 (L1): review fix stage by @11me in #46
• M10.1 (L3a): review loop driver by @11me in #47
• docs: add backlog of deferred work by @11me in #48
• M10.2: live agent output for execute/review/fix/loop by @11me in #49
• M10.3 (L3b): review loop stop conditions (stalemate + K-consecutive-clean) by @11me in #50
• M11.0: clarify suggest (agent proposes clarification questions) by @11me in #51
• M11.1: contract drafting (agent drafts contract from answers) by @11me in #52
• M11.2: Dedup agent-run attempt lifecycle into a shared helper by @11me in #53
• docs: mark lifecycle dedup resolved (#53) by @11me in #54
• M11.3: Remove dead code and add a deadcode gate by @11me in #55
• M11.4: Fix review-loop open_findings count + dedup re-proposed findings by @11me in #56
• M11.5: Mechanical path-scope enforcement in the gate (advisory) by @11me in #57
• M11.6: Idle (no-output) agent timeout instead of total wall-clock by @11me in #58
• M11.7: Run the race detector in CI by @11me in #59
• M11.8: Distinct gate_failed terminal when a fixer breaks the gate by @11me in #60
• M11.9: Version-control the durable run record, not the transcripts by @11me in #61
• M11.10: Project map scan honors the repo's .gitignore by @11me in #62
• M11.11: Make path-scope enforcement blocking by default by @11me in #63
• M12.0: Token accounting — slice 1 (write-stage usage) by @11me in #64
• M12.1: Token accounting — slice 2 (read-stage usage) by @11me in #65
• M12.2: Token budget stop for the review loop (max_tokens) by @11me in #66
• M12.3: Version pactum's own .heurema durable run record by @11me in #67
• M12.4: Cross-model review panel — concurrent N-reviewer rounds by @11me in #68
• M12.5: Split the oversized app.go into cohesive same-package files by @11me in #69
• M12.6: Introduce a Store port for the workspace durable record by @11me in #70
• M12.7: Raise the default review-loop max-rounds cap to 10 by @11me in #71
• M12.8: Review-loop convergence slice 1 — resolve-on-fix by @11me in #72
• M12.9: Review-loop convergence slice 2 — severity-gated convergence by @11me in #73
• M13.0: Add pactum usage --all (workspace cross-run token aggregate) by @11me in #74
• M13.1: Introduce an agents.Transport seam for agent execution by @11me in #75
• M13.2: Add an ACP agent transport (coder/acp-go-sdk) by @11me in #76
• M13.3: ACP transport config opt-in + agents.transport docs by @11me in #77
• Remove stray docs/acp-smoke.md smoke-test artifact by @11me in #78
• M13.5: Enforce contract path-scope at the ACP write boundary by @11me in #79
• M13.6: Harden the ACP transport — usage parity + cross-platform cleanup by @11me in #80
• audit: record dogfood run-records (M12.4–M13.6) by @11me in #81
• M14.0: De-duplicate the agents transport attempt plumbing by @11me in #82
• M14.1: De-duplicate the run-context loading prefix in internal/app by @11me in #83
• M14.2: Small cleanups — fold reviewLoopLimits into reviewLimits, drop dead RunResult.Command/Args by @11me in #84
• audit: record dogfood run-records (M14.0–M14.2) by @11me in #85
• M14.3: Stop the ledger leaking absolute repo_root; fix stale AGENTS.md by @11me in #86
• audit: scrub leaked repo_root from committed ledger + record M14.3 run by @11me in #87
• M15.0: Clarification — recommended answers + explore-first (grill-me slice 1) by @11me in #88
• M15.1: Stop codexAgentMessageText gluing successive agent messages by @11me in #89
• docs: backlog — investigate committed .heurema run-record growth by @11me in #90
• M15.2: Clarification — dependency-ordered questioning (grill-me slice 2) by @11me in #91
• M15.3: Clarification — terminology / domain...