v0.1.11 - Benchmark tasks, Rust profile, and localization coverage

Patch release for deterministic benchmark tasks, stack-profile fixture coverage, and README localization drift hardening. This release adds repository-owned benchmark task specs for the external runner while preserving the rule that project-specific oracles live in this repository, not in the runner.

Added

• Rust profile guidance, fixture coverage, smoke-test wiring, installer coverage, and README/profile documentation so Rust crate and Cargo workspace targets have a conservative local verification path.
• Buildable gobasic package in the go-basic fixture plus a Go toolchain smoke test that runs the installed Go profile check_harness.py (go build, go vet, go test) when go is available and skips otherwise, closing the verification gap from issue #41.
• Eight deterministic benchmark task definitions under benchmarks/tasks/, covering small bugfixes, docs-only boundaries, forbidden-file guards, failure memory, decision memory, profile scope, installer safety, and command workflow guidance.
• Benchmark documentation and task outcome evidence for runner smoke checks, Codex dry-run oracle fixes, and benchmark ownership boundaries.
• Traditional Chinese README localization and README language-switcher drift coverage.

Changed

• Harden benchmark task tests so each task has a deterministic oracle, narrow expected files, explicit forbidden files, and required expected-file edits where the runner treats expected_files as an allowlist.
• Normalize Markdown-oriented benchmark oracles for docs-only and refresh workflow tasks so ordinary line wrapping does not create false negatives.
• Refresh README badges, localized README image references, repository-transfer URLs, static-site metadata, profile lists, and validation docs.
• Update /harness update and adoption prompt guidance around source tracking and current repository URLs.

Validation

• python3 -m unittest discover -s tests (194 tests, 2 skipped)
• python3 -m py_compile scripts/apply_harness.py scripts/check_docs_drift.py scripts/check_structure.py scripts/check_encoding_hygiene.py scripts/check_effectiveness_plan.py scripts/check_failure_memory.py scripts/check_decision_memory.py scripts/harness_doctor.py
• python3 scripts/check_docs_drift.py
• python3 scripts/check_structure.py
• python3 scripts/check_encoding_hygiene.py
• python3 scripts/check_effectiveness_plan.py
• python3 scripts/check_failure_memory.py
• python3 scripts/check_decision_memory.py
• python3 scripts/harness_doctor.py --target . (100/100)
• git diff --check

v0.1.10 - Harness Doctor v2 coupling diagnostics

Patch release for Harness Doctor v2. This release turns Doctor from a flat baseline scan into a six-element repository health and coupling diagnostic while preserving the boundary between harness readiness and agent effectiveness.

Added

• Harness Doctor v2 scoring across Instructions, Constraints, Feedback, Memory, Evaluation, and Governance.
• First-class coupling findings for orphan constraints, orphan feedback, unoperationalized memory, unevaluated memory, ungoverned change types, and promotion gaps.
• Optional Doctor gates for minimum score and critical coupling findings, disabled by default.
• Decision memory for the Doctor v2 model and task outcome evidence for the implementation and review loop.

Changed

• Update /harness doctor, the scoring rubric, example reports, component map, theory docs, and roadmap guidance to describe the six-element diagnostic.
• Keep Proven/effectiveness signals unmeasured in Doctor output unless durable outcome evidence supports a claim.
• Tighten feedback-binding heuristics so generic documentation mentions do not count as execution wiring, and unbound check scripts do not inflate Feedback health.
• Expand Doctor regression tests for coupling findings, optional gates, non-comparable outcome evidence, illustrative effectiveness reports, and feedback-binding edge cases.

v0.1.9 - Operational evidence and command-reference validation

Patch release for operational evidence tracking, Go profile coverage, and command-reference validation. This release strengthens the kit's ability to collect trustworthy agent-work evidence without turning the starter kit into a heavier automation framework.

Added

• Go profile guidance, fixture coverage, smoke-test wiring, and README/profile documentation so Go targets have a conservative local verification path.
• Task outcome evidence decision guidance for substantial harness work, including required evidence fields for included task outcome records.
• Dogfood and effectiveness evidence reports for Today Bus, Harness ERP, and small evidence-pass scenarios, plus task outcome examples for harness adoption and maintenance work.
• Failure records and decision memory for dogfood evidence consistency and first-pass task outcome evidence gaps.
• Google site verification, sitemap, robots, and static-site metadata updates.

Changed

• Extend scripts/check_effectiveness_plan.py, plus the generic template copy, to validate task outcome evidence fields and reject contradictory or stale effectiveness-report completion language.
• Extend scripts/check_failure_memory.py and scripts/check_effectiveness_plan.py, plus generic template copies, to validate root make targets and root just recipes referenced by failure-memory records, adoption reports, and task outcome verification commands.
• Strengthen dogfood evidence validation, effectiveness templates, adoption evidence checklists, roadmap guidance, and evaluation docs around operational evidence loops.
• Refresh README, localized README files, contributor visuals, profile lists, validation docs, and lifecycle pilot notes to match the current evidence and profile coverage.
• Revert the Crowdin localization sync path while preserving localized README consistency.

Validation

• python3 -m unittest discover -s tests (146 tests, 1 skipped)
• python3 -m py_compile scripts/apply_harness.py scripts/check_docs_drift.py scripts/check_structure.py scripts/check_encoding_hygiene.py scripts/check_effectiveness_plan.py scripts/check_failure_memory.py scripts/check_decision_memory.py scripts/harness_doctor.py
• python3 scripts/check_docs_drift.py
• python3 scripts/check_structure.py
• python3 scripts/check_encoding_hygiene.py
• python3 scripts/check_effectiveness_plan.py
• python3 scripts/check_failure_memory.py
• python3 scripts/check_decision_memory.py
• python3 scripts/harness_doctor.py --target . (98/100)
• git diff --check

v0.1.8

Summary

• Hardened failure-memory verification so failure records must point to concrete detection or prevention evidence instead of non-committal future checks.
• Added scripts/check_failure_memory.py and kept the generic template copy aligned.
• Extended adoption and effectiveness report checks to validate failure-memory linkage, concrete path references, and failure-memory fields.
• Added root package.json script existence validation for npm, pnpm, yarn, and bun run <script> references.
• Documented the Today Bus Next.js dogfood target alongside the Django dogfood target.

Validation

• python3 -m unittest discover -s tests -> 115 OK, 1 skipped
• python3 -m py_compile scripts/apply_harness.py scripts/check_docs_drift.py scripts/check_structure.py scripts/check_encoding_hygiene.py scripts/check_effectiveness_plan.py scripts/check_failure_memory.py scripts/check_decision_memory.py scripts/harness_doctor.py
• python3 scripts/check_docs_drift.py
• python3 scripts/check_structure.py
• python3 scripts/check_encoding_hygiene.py
• python3 scripts/check_effectiveness_plan.py
• python3 scripts/check_failure_memory.py
• python3 scripts/check_decision_memory.py
• python3 scripts/harness_doctor.py --target . -> 100/100
• git diff --check
• git diff --cached --check

v0.1.7

Summary

• Added deterministic behavior check gate-placement guidance across harness review, refresh, adoption, generic AGENTS, and verification checklist workflows.
• Added ADR and failure memory for deterministic product-behavior checks that remain focused/manual without gate-placement review.
• Added adoption report gate-placement fields and examples for normal, focused, and manual verification paths.
• Extended check_effectiveness_plan.py and the generic template copy to validate adoption report gate-placement fields, including exact heading matching and wrapped/nested field values.

Validation

• python3 -m unittest discover -s tests -> 77 OK, 1 skipped
• python3 -m py_compile scripts/apply_harness.py scripts/check_docs_drift.py scripts/check_structure.py scripts/check_encoding_hygiene.py scripts/check_effectiveness_plan.py scripts/check_decision_memory.py scripts/harness_doctor.py templates/generic/scripts/check_effectiveness_plan.py
• python3 scripts/check_docs_drift.py
• python3 scripts/check_structure.py
• python3 scripts/check_encoding_hygiene.py
• python3 scripts/check_effectiveness_plan.py
• python3 scripts/check_decision_memory.py
• python3 scripts/harness_doctor.py --target . -> 100/100
• git diff --check
• git diff --cached --check

v0.1.5

Patch release for the decision-memory follow-up to v0.1.4. This release moves the decision-docs gate from review-only guidance into the generic target template that future adoptions copy.

Added

• Decision-memory guidance in the generic AGENTS.md template for non-trivial product or workflow structure, integration or mock external-behavior boundaries, major data models, state classifications, or UX principles that become code structure.
• A completion criterion requiring agents to report whether decision docs were added, an existing ADR covers the choice, or no decision record was needed.
• A Decision-docs gate field in the harness review report template so the specific /harness review diagnostic does not get lost when reviewers use the template.
• Regression coverage that keeps the generic completion gate and review report gate wired in.

Validation

• python3 -m unittest discover -s tests
• python3 -m py_compile scripts/apply_harness.py scripts/check_docs_drift.py scripts/check_structure.py scripts/check_encoding_hygiene.py scripts/check_effectiveness_plan.py scripts/harness_doctor.py
• python3 scripts/check_docs_drift.py
• python3 scripts/check_structure.py
• python3 scripts/check_encoding_hygiene.py
• python3 scripts/check_effectiveness_plan.py
• python3 scripts/harness_doctor.py --target .

v0.1.4

Patch release for governance documentation and review diagnostics. This release keeps /harness review diagnostic-only while tightening the durable-memory review path and clarifying command usage.

Added

• Failure memory for missed ADR review when structural product decisions are implemented without decision-record consideration.
• A /harness review diagnostic warning for product or workflow structure, mock external-behavior boundaries, major data models, state classifications, or product UX principles that become code structure without a docs/decisions/ update or explicit justification.
• iOS as a roadmap candidate profile paired with the existing Android profile, with Xcode, simulator, signing, and device checks documented as macOS/manual unless a target repository already has macOS CI.

Changed

• Clarify that /harness ... names are prompt conventions by default, not built-in editor commands.
• Refine /harness review sub-agent ownership guidance so reviewer mode and fallback reason stay parent/orchestrator-owned.
• Update localized README command guidance to match the English prompt convention wording.

Validation

• python3 -m unittest discover -s tests
• python3 -m py_compile scripts/apply_harness.py scripts/check_docs_drift.py scripts/check_structure.py scripts/check_encoding_hygiene.py scripts/check_effectiveness_plan.py scripts/harness_doctor.py
• python3 scripts/check_docs_drift.py
• python3 scripts/check_structure.py
• python3 scripts/check_encoding_hygiene.py
• python3 scripts/check_effectiveness_plan.py
• python3 scripts/harness_doctor.py --target .

v0.1.3 - Harness review sub-agent routing fix

Patch release for /harness review reviewer-mode routing. This keeps the command diagnostic-only while making subagent availability and fallback reporting harder to skip silently.

Added

• Explicit /harness review sub-agent invocation mode for read-only reviewer subagent use when the active runtime and tool instructions allow it.
• Review report Invocation, Reviewer mode, and Fallback reason fields in the template and example report.
• Regression coverage for subagent fallback guidance, prompt drift, localized README wiring, and route precedence.

Changed

• Clarified /harness review fallback behavior when a subagent tool is present but not permitted by active tool instructions.
• Routed /harness review sub-agent before the generic /harness review command in agent-facing prompts and command routing.

Validation

• python3 -m unittest discover -s tests
• python3 -m py_compile scripts/apply_harness.py scripts/check_docs_drift.py scripts/check_structure.py scripts/check_encoding_hygiene.py scripts/check_effectiveness_plan.py scripts/harness_doctor.py
• python3 scripts/check_docs_drift.py
• python3 scripts/check_structure.py
• python3 scripts/check_encoding_hygiene.py
• python3 scripts/check_effectiveness_plan.py
• python3 scripts/harness_doctor.py --target .

v0.1.2

Governance release for change-set review. This release adds a diagnostic
/harness review workflow so maintainers can challenge current changes before
completion without adding runtime hooks, policy enforcement, CI adapters, or
more automatic installer behavior.

Added

• /harness review command workflow for opposing harness-engineering review of
  the current change set.
• Harness review report template and example report.
• Quick Start, full adoption prompt, localized README, static site, and
  component-map wiring for /harness review.
• Regression tests that keep /harness review command routing, localized docs,
  report-template sections, and prompt drift covered.

Changed

• Clarify that the existing harness review checklist is a periodic maintenance
  checklist, distinct from the /harness review change-set command.
• Update the roadmap from adding /harness review to refining it through real
  target-repository use.

v0.1.1

Stabilization release for the initial harness workflow. This release strengthens
the theory, evaluation, failure-memory, and contributor guidance added around
the v0.1.0 early release.

Added

• Harness engineering theory document that separates repository harness health
  from observed agent effectiveness.
• Task outcome record template for comparable agent-work observations.
• Roadmap and expanded contributor guidance for profiles, drift checks,
  adoption examples, and release validation.
• Regression coverage that keeps the static site copy prompt aligned with the
  README adoption prompt.

Changed

• Compact root and generic AGENTS.md guidance while preserving command
  routing, analysis, validation, and commit rules.
• Clarify python3 validation commands for macOS/Linux environments where
  python is unavailable.
• Clarify Harness Doctor score scope and non-scored evaluation/governance
  signals.
• Strengthen adoption and update guidance around failure-memory records for
  user-visible runtime failures, high-risk bug paths, failed checks, repeated
  agent mistakes, and cross-environment mismatches.