Skip to content

CHANGELOG.md

Latest commit

 

History

History
391 lines (268 loc) · 56.7 KB

CHANGELOG.md

File metadata and controls

391 lines (268 loc) · 56.7 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

Please choose versions by Semantic Versioning.

  • MAJOR version when you make incompatible API changes,
  • MINOR version when you add functionality in a backwards-compatible manner, and
  • PATCH version when you make backwards-compatible bug fixes.

v0.32.1

  • refactor(changelog): convert rule changelog/unreleased-entry-required from judgment to script tier — deterministic check in scripts/rule-checks.sh (state-check on the PR-HEAD ## Unreleased section) instead of model judgment. Fires reliably even on a weak review model (the maintainer bot's MiniMax reviewer was silently skipping the judgment funnel). docs/changelog-guide.md Enforcement line rewritten (→ enforcement_type: script), @commits Trigger dropped, rules/index.json regenerated. Trade-off: no trivial-diff waiver.

v0.32.0

  • feat: Add review rule changelog/unreleased-entry-required (SHOULD, judgment, @commits trigger) in docs/changelog-guide.md — flags any PR that changes non-vendored source in a repo with a CHANGELOG.md but adds no ## Unreleased bullet. Maps to request-changes, so the maintainer bot won't auto-merge a PR that would leave the release agent nothing to ship. Feeds /coding:pr-review, /coding:code-review, and the maintainer bot (which inlines pr-review.md) from the single rule index.

v0.31.0

  • feat: /coding:self-improve Step 5 becomes an adversarial placement procedure — (5a) mandatory discovery (semantic search + Glob/Grep) that must name 2–3 concrete candidate homes before any routing, with query discipline (search the workflow/domain name, not the rule text; guides often live outside the repo) and a no-forfeit rule (empty discovery never routes to a CLAUDE.md — a memory target must beat a named alternative), (5b) a 5-rung placement ladder (repair misfiring artifact → extend point-of-mistake artifact → point-of-use guide/runbook → project CLAUDE.md → global CLAUDE.md) where descending requires naming the rung's STRONGEST discovery candidate — quoting a heading from the guide/runbook candidate to prove it was opened — and why it can't own the fix, and (5c) a 4-part global-CLAUDE.md admission argument (always-on with dead-weight counter-test, unowned, behavioral-not-procedural, one stable line) plus a de-generalization test (a rule whose un-generalized wording names a workflow step is procedural — rewording it to sound behavioral is the tell of a misroute) with auto-demotion to the best discovered home when any argument fails; Step 2 records candidates as incidents and forbids drafting rule wording before placement; output's Artifacts checked: line replaced by a Placement: ladder trace; adds mcp__semantic-search__search_related to allowed-tools. Hardens the v0.30.0 artifact-first routing after a session where a PR-workflow rule was still proposed for global CLAUDE.md despite a domain guide with exact neighboring bullets being one search away.

v0.30.2

  • fix: migrate MacPorts paths to Homebrew across shellcheck-assistant (/opt/local/bin/shellcheck → /opt/homebrew/bin/shellcheck in allowed-tools + docs, drop port install), license-assistant (gdate for current year), and the toolchain-preflight/scaling/findings scenarios (ast-grep → /opt/homebrew/bin/ast-grep) — MacPorts was removed in the local MacPorts→Homebrew migration.

v0.30.1

  • fix: checker templates exclude no-fix advisory GO-2026-5932 (golang.org/x/crypto/openpgp, unmaintained) — add it to VULNCHECK_IGNORE in Makefile.library/Makefile.service, and introduce a shared TRIVY_IGNORE baseline merged with any repo-local .trivyignore via a temp ignorefile, and align Makefile.service's trivy target with Makefile.library's full flag set (--db-repository, --ignorefile, --secret-config, --skip-dirs vendor). Propagates the exclusion to all consuming repos across both scanners.

v0.30.0

  • feat: rewrite /coding:self-improve Step 5 routing to artifact-first, CLAUDE.md-last — routing table now leads with existing artifacts (command/agent/skill repair, runbooks, guides) and demotes memory files to last resort with a mandatory "which artifacts were checked" justification; adds a −1 scoring signal for proposals that route to a CLAUDE.md an existing artifact could own. Counters the observed bias of every session proposing new memory rules.
  • fix: /coding:self-improve audit follow-ups — add allowed-tools frontmatter (repo convention, 10/16 commands declare it), generalize Phase 2 apply-style rules beyond CLAUDE.md to all target artifact types, add mandatory Artifacts checked: line to the per-proposal output template, annotate the −1 scoring signal as judged-after-routing, fold the repair-before-route-around paragraph into the last-resort paragraph, refresh the stale frontmatter description.

v0.29.0

  • feat: add a worth-it rating rubric to /coding:self-improve Step 3 — gate-then-grade (hard evidence gate + weighted score: recurrence/cost/generality/repair, minus obvious-anyway) with propose/borderline/discard tiers, mirroring /vault-cli:reflect's significance filter; surfaces a Worth-it: score per proposal so borderline candidates get an inspectable number instead of a coin-flip.

v0.28.2

  • docs: add "Repair before route-around" rule to /coding:self-improve — when an existing command/agent/skill misfires, fix that artifact instead of adding a CLAUDE.md workaround rule; only fall back to a memory rule when the tool genuinely can't be changed.

v0.28.1

  • docs: add "Version Alignment Is Release-Time, Not Feature-PR-Time" section to docs/changelog-guide.md — tells the PR reviewer that in autoRelease repos a feature PR keeps ## Unreleased and does NOT bump the manifest version strings (release agent aligns them post-merge). Prevents the recurring false-positive "version alignment violation" the reviewer raised on every autoRelease feature PR that adds an Unreleased bullet. Adds a matching validation checklist item.

v0.28.0

  • feat: add /coding:self-improve command — reviews the current session and proposes at most two durable, evidence-backed improvements to the Claude Code environment (memory/CLAUDE.md rules, commands, agents, skills). Two-phase: Report (read-only, ≤2 proposals ranked, routes each by scope) → Apply (only after explicit approval). Default outcome is "nothing worth keeping this session." Stays inline (analyzes the parent conversation; cannot be delegated to a sub-agent). Sibling to the audit-* commands, on the friction-removal side.

v0.27.0

  • feat: add factory rule go-factory/main-holds-only-boot-lifecycle-config (SHOULD) to docs/go-factory-pattern.md. New §11 "The main.go / factory boundary" states the complementary positive rule to §6.2/§8: main.go Run keeps only boot errors, lifecycle (defer), and config branching — every pure-composition value (including run.Func server/consumer wiring) belongs in a factory. Adds the "pass-through test", tightens §4.3 ("no additional logic" limits statement kind, not count — multi-statement router wiring is fine), fixes the §8 example which previously modeled router assembly in main, and adds a Summary bullet. Regenerates rules/index.json (judgment-tier, trigger **/main.go) and surfaces the rule in go-factory-pattern-assistant.

v0.26.0

  • feat: split review commands into 3 distinct scopes. Renames current /coding:code-review → /coding:local-review (preserving its diff-vs-HEAD~1 semantics for pre-commit local checks), AND introduces a brand-new /coding:code-review that scans the WHOLE codebase via git ls-files. New command ships with severity filter (default-on: Must Fix + Should Fix only; --include-optional to opt in), rule-id dedup at consolidation (N occurrences → 1 summary with sample sites), and baseline file (.code-review-baseline.yaml via --refresh-baseline) so subsequent runs only flag NEW findings (drift since last sweep) rather than the operator's full accepted tech-debt set. Design rationale: docs/three-command-review-split.md. 10 reference sites updated across llms.txt, README.md, scenarios/*, and 4 agent definitions to point at /coding:local-review (preserving old semantics). Migration: sharp behavior cutover on the /coding:code-review slot — operators previously relying on its diff-vs-HEAD~1 behavior must move to /coding:local-review. The contrast pair pr-review (remote, branch vs target) / local-review (local, pre-commit) reads cleanly; code-review takes the unmarked whole-codebase slot.

v0.25.1

  • docs: add docs/three-command-review-split.md — design note locking the proposed shape for splitting the review commands into three distinct scopes (/coding:pr-review remote diff, /coding:local-review local pre-commit diff (renamed from current /coding:code-review), new /coding:code-review whole-codebase audit). The doc pins defaults for baseline-file location (.code-review-baseline.yaml at repo root, --baseline-path override), monorepo handling (one baseline per repo by default; per-subproject via override), --refresh-baseline clean-tree requirement, and .gitignore semantics (respected; no separate .claude-ignore introduced).

v0.25.0

  • feat: extract coding:architecture-dimensions-assistant agent + docs/architecture-dimensions-guide.md — closes the v0.24.0 /coding:architecture-review MAJOR finding ("dimensions pass routes to generic claude with inline agent role"). The 8-dimension behavioral review checklist (data flow, failure, concurrency, observability, cross-cutting consistency, config/blast radius, evolvability, drift) is now a maintainable doc-paired agent. /coding:architecture-review Agent B now routes to the new agent with a one-line prompt instead of an inlined 200-word checklist. Indexed in README.md (Workflows & Documentation + Go agents table), llms.txt, and CLAUDE.md Doc↔Agent table.

v0.24.0

  • feat: add /coding:architecture-review [directory] — deep whole-codebase architectural review (different altitude from /coding:code-review's diff-scope and /coding:audit-architecture's single-agent scan). Spawns two parallel agents: top-down structural (go-architecture-assistant / python-architecture-assistant) + dimensions pass (data flow, failure, concurrency, observability, drift). Consolidates into Must Fix / Should Fix / Could Fix with file:line citations and Top 5 highest-leverage fixes.

v0.23.1

  • docs: add docs/go-package-layout-guide.md — flat pkg/ default, five subpackage-split triggers (independent reuse, versioning, cycle break, navigation friction, build-tag isolation), conventional always-split exceptions (pkg/factory/, pkg/handler/), GOOD/BAD code pair, premature-split → flat migration steps. Indexed in README.md (Go — Architecture & Patterns), llms.txt, and CLAUDE.md Doc↔Agent table (owner: go-architecture-assistant).

v0.23.0

  • feat: add RULE go-composition/no-same-package-private-helper-for-business-logic (MUST, judgment-tier, owner go-architecture-assistant) — flags new business logic added as a same-package private helper instead of behind an interface + constructor + struct + method seam. Closes the gap left by the cross-package version (which only matches pkg.Func syntax). Codified after PR bborbe/recurring-task-creator#16 shipped a private buildFrontmatter helper that local pr-review dismissed as "pre-existing pattern" (broken-windows fallacy).

v0.22.0

  • feat!: selector mode is now the DEFAULT for /coding:pr-review and /coding:code-review (was opt-in via --selector/selector token); callers passing standard now get selector behavior — BREAKING CHANGE for any tooling that relied on standard triggering per-owner Task dispatch
  • feat!: standard-mode per-owner Task dispatch removed — owners_to_spawn computation, per-owner Task prompt template, funnel clean — no adjudication needed standard-mode short-circuit, and REVIEW_TIMING per-owner instrumentation block deleted from both commands; full mode remains the per-owner deep sweep escape hatch
  • docs: scenarios/002 and scenarios/004 retired (status: outdated) — superseded by scenarios/005 and scenarios/006 respectively; footer updated with retirement reason
  • chore: acceptance.sh updated to assert the new dispatch reality: Selector mode (default) routing check added (section 1), per-owner block check tightened to full-mode-only assertions (section 2)

v0.21.0

  • fix: ast-grep-runner.sh excludes .git/ paths from both input file list and findings output — incident 2026-06-11 surfaced 274 stale agent-auditor findings from .git/COMMIT_EDITMSG during selector validation
  • feat: scenarios/005 (selector clean short-circuit) and scenarios/006 (selector findings path) — two new E2E scenarios (walked 2026-06-11, promoted to active) contracting the selector-mode journeys for a README-only diff and the perpetual fixture PR
  • feat: acceptance.sh section 5 "Selector mode contracts" — 16 scripted checks covering --selector parse, short-circuit string, GUIDE_OK/GUIDE_MISSING fail-fast, selector-mode-guide.md content, sibling consistency, citation-validator rejection of unknown rule_ids, and runner .git/ exclusion

v0.20.0

  • feat: add opt-in --selector mode to /coding:pr-review and /coding:code-review — replaces per-owner Task fanout (Step 4b-ii) with two in-session steps: Step 4c-sel CLASSIFY (narrows glob candidates via recall-optimized contract) and Step 4d-sel ADJUDICATE (reads only applicable rule blocks, emits findings in existing severity buckets); zero sub-agent spawns; default mode unchanged
  • refactor: extract selector-mode classify/adjudicate procedure to docs/selector-mode-guide.md — commands carry thin pointers (single source of truth; answers pr-reviewer command-thin + single-source-of-truth findings)
  • fix: selector-mode guide resolution is now fail-fast — explicit GUIDE_OK/GUIDE_MISSING echo, and on missing guide the review STOPS with a Must Fix toolchain failure instead of silently continuing mechanical-only (caught by MiniMax-M2.7 benchmark: weaker model skipped classify+adjudicate entirely when the guide path failed, presenting a judgment-less review as complete)

v0.19.0

  • refactor!: BREAKING CHANGE — rename coding:release-changelog-agent → coding:release-changelog-assistant to match marketplace <noun>-<role>.md naming convention (-assistant / -auditor, not -agent). Coordinated rename: agent file agents/release-changelog-agent.md → agents/release-changelog-assistant.md, name: frontmatter, both callers' subagent_type= reference, README agents-table entry, llms.txt entry, and CHANGELOG mentions. Callers on v0.18.0 will fail to resolve until updated; v0.18.0 callers are bundled in this same commit so any installer of v0.19.0+ has the matching pair.
  • refactor: restructure coding:release-changelog-assistant body to XML schema per agent-command-development-guide — add <constraints> block (NEVER commit/tag/push, ALWAYS return error field when malformed, etc.), <process> block summarizing the 5-step workflow up-front (was buried), <output_format> for the success JSON schema, <error_handling> for the three error codes (separates concerns from success). Markdown # headings retained only for high-level sections (Purpose, Inputs, classification rules, rewrite rules, caller profiles, invocation example); structural directives now use XML tags as the guide requires.
  • docs: trim coding:release-changelog-assistant description: frontmatter from 229-char paragraph to one-sentence summary per agent-command-development-guide; the long-form caller list + flag semantics live in the body, not the description.
  • chore: drop non-standard effort: medium frontmatter from coding:release-changelog-assistant — not a documented Claude Code field; if smaller-model cost is wanted, use model: haiku (documented mechanism).
  • docs: add coding:release-changelog-assistant entry to llms.txt under Development Workflows so plugin installers + /coding:check-guides discover it alongside the changelog guide.

v0.18.0

  • fix: address third round of pr-reviewer findings — add missing coding: plugin namespace prefix to subagent_type in 3 places (commit.md, github-release.md, agents/release-changelog-agent.md invocation example); agent would have failed to resolve at runtime when installed via marketplace. Strip pre-release suffix in version arithmetic (v0.69.0-rc1 → 0.69.0 before read MAJ MIN PAT) so the command's bash matches the agent's documented input contract. Drop unused Bash(sed:*) from allowed-tools (swapped to awk earlier). Replace remaining ~/Documents/workspaces/maintainer example with /path/to/local/repo (self-containment).
  • refactor: /coding:github-release now invokes release-changelog-agent for bump classification + bullet rewrite (flags: majorBumpAllowed=true, rewriteChangelogEntries=true). Step 4 replaces the inline classifier prose; Step 8 threads the optional rewritten_unreleased body through to the header rewrite (Edit tool for full-block replacement, awk for header-only passthrough). Operator confirm at Step 6 still has final say on the agent's bump call (downgrade dialog).
  • refactor: /coding:commit Workflow B now invokes release-changelog-agent for bump classification instead of using the inline Version Increment Rules heuristic. Flags: majorBumpAllowed=false (preserves Workflow B's "major requires manual edit" contract), rewriteChangelogEntries=false (pure passthrough — Workflow B keeps the sed-rename behavior, no AI rewrite). Behavior on a happy-path release is identical to pre-refactor; the change is internal (one canonical source of bump-classification rules for all three callers).
  • refactor: release-changelog-agent now reads CHANGELOG.md from cwd and extracts the ## Unreleased block itself (instead of requiring the caller to pass unreleased_body inline). Centralizes parsing logic in one place; callers only pass current_version + the two flags. New output field unreleased_body returns the verbatim extracted text so the caller can do an atomic Edit-tool match-and-replace without re-parsing. Adds an error-case JSON branch (changelog-missing / unreleased-section-missing / unreleased-section-empty) for malformed input.
  • feat: add release-changelog-agent sub-agent — canonical release-AI source: classifies the next semver bump from ## Unreleased bullets and optionally rewrites them to conventional-prefix style. Two-flag contract (majorBumpAllowed, rewriteChangelogEntries) lets each caller opt into the capability tier it wants. Plan-time only (no git mutation). Ported from agent/github-releaser/pkg/prompts/bump_classification.md + changelog_rewrite.md with the flag layer added. Faithfulness audit (the agent's ai_review phase) deferred to Phase 2.
  • feat: add /coding:github-release command — direct release of a git repo (cwd, local dir, or owner/repo clone-to-tmp), with bump classification from ## Unreleased, commit/tag/push, and PR + auto-merge fallback for branch-protected repos
  • fix: address second round of /coding:github-release pr-reviewer findings — define owner_repo helper for Step 7's tag-collision gh api call (previously referenced undefined $OWNER/$REPO), add trap '... rm -rf ...' EXIT to clean tmp clones on any exit path (including Ctrl-C and errors), spell out Step 11's PR-merge polling loop (for _ in $(seq 1 30); do sleep 10; ...) including the re-run path on timeout, drop personal-path references from the command body + CHANGELOG bullet (marketplace self-containment), add new command to README.md commands table
  • fix: address /coding:github-release pr-reviewer findings — quote argument-hint, tighten allowed-tools rm -rf glob to mktemp-pattern (/tmp/github-release/tmp.*), reject owner/repo targets containing : or @ (host-injection), swap sed→awk in CHANGELOG header rewrite (metachar-safe), define die + default_branch helper functions inline
  • refactor: scope /coding:github-release tmp clones to /tmp/github-release/ (was bare mktemp -d landing in $TMPDIR — /var/folders/... on macOS, /tmp on Linux); tightens allowed-tools from rm -rf /tmp:* + rm -rf /var:* (overly broad, allowed rm -rf /var system dir) to rm -rf /tmp/github-release:* only

v0.17.0

  • fix: correct Step 4b-i trigger glob-to-regex conversion in both commands — escape literal dots, anchor the match, **/ matches zero-or-more dirs (root main.go), placeholder ordering so ** expansion survives the * pass; found by self-reviewing PR #48 with its own pipeline (12 phantom owners → 1 real)

  • perf: add standard-mode early exit to Step 4 in both commands — diffs touching no rule-relevant files (no .go/.py, no CHANGELOG/go.mod/LICENSE/README/Makefile/pyproject/k8s/agents/commands/skills/docs) skip the funnel entirely with a report note

  • docs: update scenarios/001-004 to funnel v2 architecture (scripts/ast-grep-runner.sh, exit 2, /tmp/findings.json contract, funnel-clean short-circuit, diff-scope assertions); walk 001/003/004 with real output recorded

  • feat: extend build-index.py to derive enforcement_type (mechanical/script/judgment) and parse optional **Trigger**: field into a trigger array for every rule entry in rules/index.json

  • feat: backfill **Trigger**: field for all 62 judgment-tier rules across 30 doc files — enables diff-scoped dispatcher to skip owners whose triggers don't match changed files

  • feat: rewrite Step 4 in commands/pr-review.md and commands/code-review.md — standard mode now computes active judgment-rule set via jq glob-matching and spawns only owners present in findings_by_owner ∪ active-judgment-rule-owners; zero LLM spawns when funnel is clean and no judgment rules are active; full mode unchanged

  • docs: update docs/rule-block-schema.md to document Trigger optional field and new enforcement_type/trigger index fields

  • feat: add scripts/ast-grep-runner.sh — deterministic replacement for the ast-grep-runner LLM agent; same JSON contract (stats/findings_by_owner/errors), diff-scoping via changed-file args, merges rule-checks.sh output.

  • feat: add scripts/rule-checks.sh — script-tier (bucket 2) mechanical checks for 11 rules: go-licensing/license-file-required, go-licensing/readme-license-section-required, changelog/preamble-frozen, git-commit/subject-under-50-chars, git-workflow/no-ai-attribution-in-commits, go-library/semver-vprefix-tag-required, go-tools-versioning/no-tools-go-for-clis, go-mod-replace/no-cross-repo-replace, python-project-structure/src-layout-required, python-project-structure/pyproject-toml-with-hatchling, skill-writing/scripts-in-scripts-subdir.

  • docs: deprecate agents/ast-grep-runner.md with pointer to scripts/ast-grep-runner.sh; update **Enforcement**: lines for 11 rules to cite scripts/rule-checks.sh.

  • feat: transcribe 37 judgment-tier rule recipes into mechanical ast-grep YAMLs (rules/go/ +31, new rules/python/ +6) — each rule's prose "ast-grep partial/follow-up" recipe is now a real over-inclusive first-pass filter; mechanical coverage 29 → 66 of 139 rules. go-mod-replace/no-cross-repo-replace excluded (no go.mod grammar; doc updated to agent-only enforcement).

  • feat: add ast-grep native rule tests — testConfigs: rule-tests/ in sgconfig.yml, 37 test files with ≥2 valid + ≥2 invalid snippets per rule (incl. documented exemption cases); ast-grep test -c sgconfig.yml passes 37/37.

  • docs: update 19 docs' **Enforcement**: lines to cite the new YAML paths, keeping adjudication guidance and dropping the now-redundant pattern prose.

  • perf: gate make precommit to Full mode only in both pr-review.md and code-review.md — running the full test suite is CI's job; Standard/Short mode now report "precommit skipped (standard mode) — CI covers lint+test" instead.

  • perf: make timing instrumentation opt-in via REVIEW_TIMING=1 env var in both commands — the per-Owner ts_start/ts_end + jq/grep roll-up (~14 extra tool calls) now only fires when explicitly requested; otherwise skipped entirely.

  • perf: add Step 0a-pre short-circuit to pr-review.md — if the cwd is already a clean checkout at origin/<SOURCE_BRANCH> HEAD, skip worktree creation/removal entirely (saves ~18 tool calls in the agent pod where cwd is already at PR HEAD).

  • perf: tighten ast-grep Step 4.0 preflight in both commands — run exactly one compound check, once; on failure report and skip Step 4. Explicitly forbid further investigation (which, ls rules/, retry variants) to prevent the 4-probe drift observed in prod.

  • perf: change !git diff HEAD~1 context injection to !git diff --stat HEAD~1 in code-review.md — review steps pull full diffs per-file on demand; the unconditional full diff injection was pure token waste.

v0.16.0

  • feat(rules): bootstrap docs/go-functional-composition-pattern.md with 5 ### RULE blocks (go-functional-composition/func-type-name, list-type-name, list-checks-ctx-done, list-wraps-errors-with-ctx, multi-method-func-explicit-delegate). Covers the four load-bearing conventions of the pattern (Func/List naming, ctx-aware iteration, error wrapping with caller context, multi-method nil-safe delegation).
  • feat(rules): bootstrap docs/git-commit-guide.md with 3 ### RULE blocks (git-commit/imperative-mood, git-commit/subject-under-50-chars, git-commit/feature-branch-no-tag). All judgment-tier; the agent reads commit messages via git log rather than diff-time AST patterns. The doc also carries an explicit cross-reference to the pre-existing git-workflow/no-ai-attribution-in-commits rule (canonical in docs/git-workflow.md) instead of duplicating it.
  • feat(rules): mechanical ast-grep YAML for go-testing/suite-timeout-required. Flags func TestXxx(t *testing.T) bodies that call GinkgoConfiguration() but don't assign to suiteConfig.Timeout before RunSpecs. The only cleanly-tractable of the 5 remaining go-testing/* MUST rules — the others need errcheck-equivalent type inference or file-existence checks. Mechanical YAML count: 28 → 29.
  • feat(dispatcher): per-Owner timing instrumentation in Step 4b of both commands/pr-review.md and commands/code-review.md — JSONL log per Owner agent dispatch (event / owner / findings_in / wall_ms) plus roll-up summary. Makes "is this Owner worth dispatching?" answerable with data; previously the funnel ROI was anecdotal.
  • Total rules in index after this batch: 131 → 139 (+8 — 5 functional-composition + 3 git-commit; the AI-attribution rule already existed in git-workflow.md). 0-rule docs remaining: 15 → 13 (still ~7 meta-docs of those out of scope). Mechanical YAML count: 28 → 29.

v0.15.1

  • feat(scripts): new scripts/acceptance.sh + make check-acceptance Makefile target — 12 fast assertions covering the dispatcher contract that doesn't need an E2E scenario walk: mode coverage (short/standard/full), per-Owner routing + index-to-agent integrity, Step 2.5 context-glob mappings, broken-YAML isolation. Wired into make precommit so CI catches dispatcher drift. Closes the 4 acceptance items listed on [[Refactor coding pr-review to doc-driven rules pipeline]] task page that were left over from the scenario-cut.
  • fix(rules): go-errors/no-fmt-errorf.yml rewritten as a structural rule. The original pattern: fmt.Errorf($$$ARGS) was parsed by tree-sitter Go grammar as a type_conversion_expression (because Type(arg) is a valid Go type cast at pattern-compile time), so the rule matched no real call sites — silently emitting zero findings since the YAML shipped. Replaced with kind: call_expression + structural selector_expression match on fmt.Errorf. Verified against scenario 004's fixture (pkg/scenarios-test-fixture/violations.go on bborbe/maintainer#2): the Boom function's fmt.Errorf now fires. Scenario 004's findings_count floor lifted ≥4 → ≥5.

v0.15.0

  • feat(pipeline): dispatcher refactor for /coding:pr-review + /coding:code-review — Step 4 replaced with ast-grep-runner (mechanical funnel) → per-Owner LLM-tier adjudication → citation validation. Decouples LLM-call count from PR file count; small PR-size now equals small LLM-call count for the same rule coverage. Migrated 13 rule-enforcer agents to the dispatcher contract.
  • feat(pipeline): citation validator (scripts/validate-citations.sh) — rejects findings whose rule_id is not present in rules/index.json. Smoke-tested against synthetic hallucination payload.
  • feat(pipeline): coverage lint (scripts/check-coverage.sh) wired into make precommit — fails on dangling enforcement references, orphan YAMLs, and rule-id mismatches between docs and the index.
  • feat(dispatcher): fail-fast preflight when ast-grep / sg binary is missing — both commands/pr-review.md Step 4.0 and agents/ast-grep-runner.md Step 0 emit a documented error and exit 1 instead of silently looping on sg --version (the failure mode observed on coding#34).
  • feat(rules): mechanical ast-grep YAML count 20 → 28 across batches 5–6: go-cli/slog-not-glog-in-new-projects, go-glog/use-v-for-debug-not-info, go-testing/no-testing-t-direct, go-testing/no-stdlib-table-tests, go-architecture/constructor-returns-interface, go-architecture/no-globals-or-singletons, go-architecture/counterfeiter-directive-on-interface, go-patterns/bborbe-collection-ptr-not-helpers, go-json-error-handler/use-error-code-constants, go-k8s-binary/secret-fields-need-display-length, go-concurrency/no-raw-go-func, go-concurrency/channel-closed-by-sender-only, go-cli/cobra-not-stdlib-flag.
  • fix(rules): nosec-requires-reason.yml — pattern-regex is not a valid ast-grep 0.43 field. Rewritten as kind: comment + all: [regex: '#nosec\b', not.regex: '--']. The original YAML had been silently parse-failing on every PR review since the rule shipped.
  • feat(rules): bootstrap docs/go-logging-guide.md with 7 ### RULE blocks (no-mixing-slog-and-glog, no-log-and-return-error, external-call-logs-response, no-sensitive-data-in-logs, lowercase-log-messages, no-tight-loop-without-sampler, skip-empty-v2-heartbeats). Total rules in index: 124 → 131.
  • feat(scenarios): new scenarios/ directory with 4 active E2E acceptance scenarios following the dark-factory scenario writing guide. Each scenario is a manually-walked checklist operators run from any terminal: 001-toolchain-preflight (dispatcher fail-fasts when ast-grep is absent), 002-clean-pr-zero-findings (README-only diff emits empty severity sections), 003-scaling-funnel-100-files (mechanical funnel ≤30s, distinct Owners ≤30), 004-findings-exist-path (/coding:pr-review against the perpetual test PR bborbe/maintainer#2 surfaces ≥4 findings with valid citations).
  • docs: README + llms.txt index the 4 active scenarios alongside guides and agents.
  • fix(commit): /coding:commit no longer classifies scenarios/ changes as pipeline-only. Scenarios are shipped acceptance contracts (release-relevant, like docs/ and rules/), not dark-factory runtime state — they belong in the changelog. Prompts and specs remain pipeline-only as before. Without this fix, a PR adding only scenario files would route to Workflow E and silently ship without a release-note record (the regression observed retroactively on PR #37 + #40 + #41).

v0.14.0

  • feat(rules): grow rules/index.json from 27 to 124 entries (+97 rules across 44 doc families). Most enforceable conventions in docs/ now carry canonical ### RULE <id> (LEVEL) blocks consumable by /coding:pr-review.
  • feat(rules): new rule families bootstrapped — go-architecture, go-prometheus, go-testing, go-licensing, go-doc, go-state-machine, go-linting, go-json-error-handler, go-service-impl, go-k8s-crd, go-functional-options, go-mod-replace, go-glog, go-concurrency, go-http-service, go-cqrs, go-cli, go-enum-type, go-composition, go-build-args, go-mod-dependency-fix, go-makefile, go-patterns, go-library, go-k8s-binary, go-filter, go-parse, go-validation, go-tools-versioning, go-boolean-combinator, go-mod-dependency-fix, agent-cmd, python-architecture, python-ioc, python-logging, python-pydantic, python-project-structure, python-makefile, python-factory, tdd, test-pyramid, changelog, git-workflow, markdown-todo, claude-md, readme, skill-writing, teamvault, k8s-manifest, adr.
  • feat(rules): first mechanical ast-grep YAML for a bootstrapped rule — rules/go/counter-total-suffix.yml (enforces go-prometheus/counter-total-suffix at lint time). Recipe for struct-literal-field matching codified in docs/ast-grep-rule-writing-guide.md (PR #11 → #12).
  • feat(precommit): new make check-index target wired into the precommit chain — fails loudly when rules/index.json is stale relative to the walker output. Closes the gap that turned the PR #9 walker-regen miss into a retroactive PR #10 catch-up.
  • docs(trim): trim three oversized guides to rules-only form, moving comprehensive reference content to the maintainer's Obsidian KB so coding/docs/ stays portable for any plugin installer — go-prometheus-metrics-guide.md 2399 → 379 lines, go-testing-guide.md 1207 → 428 lines, agent-command-development-guide.md 1750 → 347 lines.
  • docs(schema): docs/rule-block-schema.md now documents the recommended **Why**: paragraph and #### Bad / #### Good example sections — closes a gap where every existing rule block carried Why but the schema doc only described the three required fields.
  • chore(rules): CLAUDE.md doc-agent alignment table grew from 13 to 44 entries to match the new rule-family coverage.

v0.13.0

  • docs(trim): trim docs/go-prometheus-metrics-guide.md 2399 → 379 lines (rules-only architecture; comprehensive reference moves to maintainer's Obsidian KB).
  • docs(trim): trim docs/go-testing-guide.md 1207 → 428 lines (same architecture).
  • docs(trim): trim docs/agent-command-development-guide.md 1750 → 347 lines with 7 pre-canonicalized ### RULE blocks (agent-cmd/command-thin, no-user-prompts, scripts-in-claude-dir, command-frontmatter, agent-frontmatter, single-source-of-truth, gap-driven-feedback).
  • feat(rules): add six ### RULE blocks to docs/go-prometheus-metrics-guide.md — go-prometheus/counter-pre-initialization, composed-metrics-interface, no-gauge-for-monotonic, counter-total-suffix, help-string-quality, label-naming-consistency.
  • feat(rules): add eight ### RULE blocks to docs/go-testing-guide.md — go-testing/no-stdlib-table-tests, no-testing-t-direct, no-bare-error-call, suite-test-file-required, main-test-with-compiles, suite-timeout-required, counterfeiter-mocks-required, libtime-injection-required.
  • feat(rules): first mechanical ast-grep YAML for go-prometheus/counter-total-suffix (rules/go/counter-total-suffix.yml) — uses pattern.context + selector + inside.stopBy + constraints.not.regex recipe; 4 TP / 0 FP across fixture cases (NewCounterVec, NewCounter, Name-position-agnostic, no false-flag on GaugeOpts/HistogramOpts).
  • docs(ast-grep): propagate struct-literal-field-matching recipe to docs/ast-grep-rule-writing-guide.md (new section + three pitfall entries + canonical-example pointer).
  • feat(precommit): add make check-index target; wired into precommit chain. Diff-based drift detection on rules/index.json — fails loud with actionable error when walker output and committed index diverge. Closes the gap exposed by the agent-cmd-guide trim's walker-regen miss (caught retroactively by the testing-bootstrap PR).
  • rules/index.json grown 17 → 42 entries across eight rule families: agent-cmd, go-context, go-errors, go-factory, go-http-handler, go-prometheus (new), go-security, go-testing (new), go-time.

v0.12.0

  • feat(rules): add four ### RULE blocks to docs/go-http-handler-refactoring-guide.md — go-http-handler/no-inline-error-handler, go-http-handler/no-inline-background-handler, go-http-handler/new-prefix-naming, go-http-handler/kebab-case-handler-files; matching ast-grep YAMLs for the two mechanical rules + rules/index.json entries grown from 17 to 21

v0.11.1

  • feat(rules): add four ### RULE blocks to docs/go-security-linting.md — go-security/file-perms-too-permissive, go-security/dir-perms-too-permissive, go-security/nosec-requires-reason, go-security/chmod-return-checked; matching ast-grep YAMLs + rules/index.json entries
  • chore(release): live-verify github-releaser-agent plugin-manifest bumping — release commit should rewrite .claude-plugin/plugin.json + .claude-plugin/marketplace.json version fields alongside the CHANGELOG (feature shipped via bborbe/maintainer PR #33).

v0.11.0

  • refactor(check-links): extract inline shell from Makefile target to scripts/check-links.sh — enables shellcheck linting of the link-check logic
  • feat(scan): add scripts/scan.sh — run ast-grep rules against external repos
  • feat(rules): add two ### RULE blocks — go-time/no-time-now-direct, go-time/no-time-time-in-fields; matching ast-grep YAMLs + rules/index.json entries
  • docs: add docs/ast-grep-rule-writing-guide.md — authoring guide for ast-grep YAML rules paired with ### RULE blocks
  • docs: add docs/go-time-injection.md — guide for the two new go-time rules
  • chore(ast-grep): add sgconfig.yml for project discovery; fix root main.go glob and expand ignores across all 3 rule YAMLs

v0.10.0

  • feat: add scripts/build-index.py — deterministic walker that extracts ### RULE blocks from docs/*.md and emits rules/index.json
  • docs: add docs/rule-block-schema.md — authoritative reference for ### RULE block contract and rules/index.json schema.
  • feat: add make build-index target — generates rules/index.json from ### RULE blocks in docs/
  • chore: commit initial rules/index.json — single entry for go-context/cancel-check-in-loop rule

v0.9.13

  • feat(commit): detect .maintainer.yaml: release.autoRelease: true and skip the tag step — github-releaser-agent owns the release. New routing rule short-circuits the branch check: master with bot opt-in routes to Workflow A (Unreleased append, no tag). See [[GitHub Auto-Release Guide]].

v0.9.12

  • refactor(auditors): slim audit-agent, audit-skill, audit-slash-command commands to thin delegators — move target-path resolution out of the command body into the agent-auditor / skill-auditor / slash-command-auditor agents (knowledge belongs in the agent), rewrite <success_criteria> from invocation-mechanics checks to outcome checks, and normalize argument-hint to [path/...] style. Behavior unchanged; addresses self-audit findings.

v0.9.11

  • docs(go): add go-k8s-binary-conventions.md guide.
  • docs(teamvault): add teamvault-conventions.md explaining the lookup-key vs raw-secret pattern; make coding:code-review aware of it.
  • docs(pr-review): mirror code-review Step 2.5 — load context-specific conventions during PR review.
  • templates: replace bare make vulncheck in Makefile.library and Makefile.service with the structured pattern from go-skeleton — adds VULNCHECK_IGNORE ?= ... allowlist of OSV IDs, runs govulncheck in JSON mode, and prints a deduplicated table (OSV id, module@version -> fixed_version, summary). Exits non-zero only on unignored findings.
  • docs(go-makefile-commands): document new make vulncheck behavior, the VULNCHECK_IGNORE and GOVULNCHECK_VERSION variables, and how to allowlist a new OSV ID per-project.
  • chore(release): version bump to 0.9.11 also forces plugin re-discovery of coding:audit-agent, coding:audit-skill, coding:audit-slash-command — fixed in 0.9.10 but never picked up because Claude Code caches command/agent discovery per plugin version (no bump = no re-scan).

v0.9.10

  • fix(auditors): unbreak coding:audit-agent, coding:audit-slash-command, coding:audit-skill commands and their corresponding agents (agent-auditor, slash-command-auditor, skill-auditor). Three slash commands were missing allowed-tools: Task in their frontmatter so the Task tool was blocked at invocation. Three agents were missing effort: high and used YAML-list tools: form (vs comma-string) which Claude Code's plugin loader silently rejected. Net: 3 previously-dead commands and 3 agents now load and function.
  • note: Claude Code caches plugin agent/command discovery by plugin version. New agents/commands added at the same version are NOT picked up by /reload-plugins — only a version bump triggers re-discovery. Document this for future plugin releases.

v0.9.9

  • docs: clarify CQRS handler-error semantics in go-cqrs.md — handler errors do NOT cause kafka replay; the result-sender wrapper emits a single Failure result and commits the offset. Use ErrCommandObjectSkipped to suppress noisy Failure results when the caller condition is non-retryable.
  • docs: Tighten CHANGELOG header rule in docs/changelog-guide.md — everything before the first ## is frozen; never insert above any header line. Prevents the failure mode where a past ## Unreleased was placed above the SemVer preamble and later release renames stranded the preamble mid-file.

v0.9.8

  • chore: extract check-versions to scripts/check-versions.sh; add make release-check (precommit + check-versions); unwire check-versions from precommit so drift during development is allowed and alignment is enforced at release time. Add docs/releasing-coding.md. Aligns with dark-factory / vault-cli / semantic-search release-gate shape.

v0.9.7

  • Re-align plugin manifests with git tag — .claude-plugin/plugin.json and .claude-plugin/marketplace.json (both metadata and plugins[0] entries) bumped from 0.8.0 → 0.9.7. Manifests had drifted across v0.9.x releases.
  • Add make check-versions precommit target — fails if CHANGELOG.md top version, plugin.json, and both marketplace.json version fields don't match. Prevents future drift.
  • Strengthen CLAUDE.md "Version Alignment — MANDATORY" section: 4-place rule, precommit enforcement, release checklist update.
  • make check-json now also validates marketplace.json syntax (was only plugin.json).

v0.9.6

  • Add docs/go-http-service-guide.md — canonical admin endpoint block (/healthz, /readiness, /metrics, /setloglevel/{level}, /gc, conditional /resetdb, /resetbucket, /trigger, /sentryalert, /testloglevel), port 9090 convention, K8s admin/port / admin/path Service annotations for gateway auto-registration, server lifecycle, security boundary, anti-patterns, validation checklist.
  • Trim /setloglevel wiring snippet from docs/go-logging-guide.md (now a one-line reference to go-http-service-guide.md); logging guide retains glog level meanings and curl examples.
  • Cross-link new guide from README.md and llms.txt under "Go — HTTP & APIs".

v0.9.5

  • Update docs/go-tools-versioning-guide.md with three lessons from the jira-task-creator migration:
    • New section "errcheck: Run via golangci-lint, Not Standalone" — recommends running errcheck via .golangci.yml instead of a standalone Makefile target. Avoids the GODEBUG=gotypesalias=1 workaround required by errcheck@v1.10.0 when the project uses Go 1.24+ generic type aliases. Includes the pattern (settings, exclusions) and migration steps.
    • "Don't downgrade during migration" note in the tools.env section — when a repo's existing tool version is newer than canonical (e.g. golangci-lint@v2.12.1 vs canonical v2.11.4), keep the newer version. Newer linters often introduce checks that catch real issues; downgrading silently regresses already-clean code.
    • Note in the migration step about indirect-dep auto-population — after rewriting a minimal go.mod and bumping direct bborbe deps to @latest, indirect deps are typically already at latest via Go's module graph resolution.

v0.9.4

  • Add docs/go-boolean-combinator-pattern.md — And / Or / Not combinator pattern for predicate-style interfaces. Captures the convention used by signalcheck and the upcoming truster package: single-method decision interface, structured result with description, slice-typed combinators, function-typed adapter, fail-safe empty-list handling, full-evaluation audit-trail default. Includes anti-patterns (naked bool returns, nesting via callbacks, vacuous truth in security contexts) and a checklist for new implementations.
  • Update docs/go-filter-pattern.md Related Patterns — replace the "Specification Pattern" placeholder with a real link to the new combinator guide.
  • Index the new guide in README.md (Go — Architecture & Patterns table) and llms.txt.

v0.9.3

  • Add templates/migrate-tools-go.sh — reusable bulk migration script for multi-module repos. Operates on current repo via git rev-parse --show-toplevel, finds all tools.go, deletes them, bumps every bborbe/* dep (direct + indirect) to @latest, runs go mod tidy, drops obsolete replaces. Idempotent. Used to migrate 128 trading sub-modules in one run.
  • Update docs/go-tools-versioning-guide.md: bump bborbe deps to @latest must include INDIRECT bborbe deps (drop the // indirect filter from the grep). Empirical example documented from backup/service (indirect bborbe/kv v1.19.4 was dragging full pollution; bumping to v1.19.6 dropped go.mod from 486 → 93 lines).
  • Update templates/prompt-migrate-tools-go.md to match the guide (script-driven @latest for direct + indirect; multi-module local-replace handling).

v0.9.2

  • Update docs/go-tools-versioning-guide.md with bborbe/* @latest recipe — script-driven dep bump replaces fragile manual version enumeration. New step in migration procedure: grep '^ github.com/bborbe/' go.mod | grep -v '// indirect\|=>' | awk '{print $1}' | xargs -I {} go get {}@latest
  • Add post-migration verification step (zero pollution check): grep -E '(cellbuf|go-header|go-diskfs|golangci-lint|osv-scanner|ginkgolinter|charmbracelet/x|denis-tingaikin)' go.mod must return zero matches
  • Add four new pitfalls based on agent multi-module migration experience: one unbumped bborbe dep brings cascade back; go mod why is the diagnostic; hardcoded version lists in prompts truncate; multi-module local-path replaces stay
  • Update templates/prompt-migrate-tools-go.md to match: explicit @latest step, post-tidy pollution check, multi-module replace handling, renumbered steps to 11

v0.9.1

  • Add templates/prompt-migrate-tools-go.md — dark-factory prompt template for migrating Go libraries from tools.go to tools.env + Makefile @version pattern. Self-contained with full migration steps, references the guide, includes verification checks. Copy to each bborbe lib's prompts/in-progress/ to drive the migration via dark-factory.

v0.9.0

  • Add docs/go-tools-versioning-guide.md covering the migration from tools.go to tools.env + Makefile @version pattern for CLI tool version pinning. Documents why tools.go is harmful (transitive dep pollution, cascade through library imports, permanent replace workarounds), the canonical tools.env source-of-truth, go run pkg@$(VERSION) invocation pattern, //go:generate directive update, the 7-step migration procedure, dependency-tree-based migration order, and pitfalls (gosec "dev" cosmetic, go run ignoring local replaces, go mod tidy -e truncation, osv-scanner v2.3.2+ broken upstream).
  • Add canonical templates/tools.env listing the version pin for every CLI tool (addlicense, counterfeiter, errcheck, ginkgo, goimports-reviser, golangci-lint, golines, go-modtool, gosec, govulncheck, osv-scanner). Pinned OSV_SCANNER_VERSION=v2.3.1 since v2.3.2+ is broken upstream.
  • Update templates/Makefile.library and templates/Makefile.service to include tools.env and use go run pkg@$(VERSION) everywhere instead of go run -mod=mod pkg. Switch golangci-lint import path to v2 (github.com/golangci/golangci-lint/v2/cmd/golangci-lint).
  • Remove templates/tools.go — superseded by the tools.env + Makefile pattern.

v0.8.0

  • Add docs/go-state-machine-pattern.md covering the phase-dispatched state machine pattern for long-running, resumable, multi-process workflows. Documents core structure (Phase enum, Status enum, Result envelope, dispatcher), Status-vs-Phase distinction with controller persistence rule, heterogeneous phases (pure-Go and external runners), external controller contract, four backward-edge patterns (Interventional Reset, Phase Unrolling, Sub-Phase loop, Circuit Breaker), parallel sub-phases via bborbe/run, anti-patterns with paired [GOOD]/[BAD] examples, and testing guidance. Generic Order/Customer/Product domain throughout. Indexed in README.md and llms.txt under Go Architecture & Patterns. Reference-only guide (no matching agent — fits CLAUDE.md exemption for pattern guides).

v0.7.1

  • Extend docs/go-build-args-guide.md with a "Vendor handling" section clarifying that go mod vendor is a build-time concern, not a precommit concern. vendor/ is gitignored in the canonical layout; Makefile.precommit's ensure target deletes any lingering vendor/ and uses -mod=mod; Makefile.docker's build target regenerates vendor just-in-time before docker build. Automation that edits Go code (dark-factory prompts, CI stages) should run go mod tidy when deps change but NEVER go mod vendor — the precommit step wipes it out immediately after. Prevents ~1–2 min of wasted time per prompt execution in repos that follow this pattern.

v0.7.0

  • Add docs/go-build-args-guide.md covering the three standard build-time injection values — BUILD_GIT_VERSION (from git describe --tags --always --dirty), BUILD_GIT_COMMIT (from git rev-parse --short HEAD), and BUILD_DATE — across all four participating files per service: Makefile.docker build-args, Dockerfile ARG/ENV/LABEL blocks (with OCI org.opencontainers.image.* labels), the Argument struct with matching env: tags, and the startup log. Integrates with the shared github.com/bborbe/metrics v0.5.0 package's BuildInfoMetrics helper — publishes a single build_info{version, commit} gauge whose value is the build timestamp in Unix seconds, with service identification via the Prometheus job label. Includes a rollout checklist and canonical PromQL queries for deployment-age alerting and rollout-state visibility.

v0.6.0

  • Add docs/go-mod-replace-guide.md covering when to use replace in go.mod (monorepo siblings yes, cross-repo no) with GOOD/BAD examples and antipatterns.
  • Add docs/k8s-manifest-guide.md covering Kubernetes manifest layout: k8s/ folder next to code, one-resource-per-file, type-suffix filename convention (-deploy.yaml, -svc.yaml, etc.), env-substituted placeholder pattern ({{ "KEY" | env }}), thin Makefile delegating to shared fragments, secret templating, and antipatterns. Aligned with github.com/bborbe/go-skeleton/k8s reference layout.
  • Expand docs/go-kubernetes-crd-controller-guide.md §2a with a naming convention section (group/kind/plural/scope rules) and rewrite §3 code generation setup: tools.go pins k8s.io/code-generator via cmd/validation-gen import, and hack/update-codegen.sh sources kube_codegen.sh from $GOMODCACHE (resolved via go list -m -f '{{.Dir}}') instead of vendor/ — since shell scripts aren't vendored by go mod vendor. Drops the go mod vendor prerequisite from the workflow.
  • Update docs/go-mocking-guide.md to require package-prefixed test suite filenames (<package>_suite_test.go) matching Ginkgo's ginkgo bootstrap default; removes bare suite_test.go references across the guide.

v0.5.0

  • Add go-architecture-assistant + python-architecture-assistant agents plus /coding:audit-architecture command. Detect naive extractions (helpers pulled out just to satisfy funlen/C901 — called once, generic names, same-file same-type), package/module boundary violations, dependency direction errors, layering leaks, and missing abstraction seams. Wired into /coding:code-review full alongside srp-checker (unit-level SRP stays there; architecture agents cover cross-unit).
  • Both agents include a Concrete patterns section with 6 (Go) / 5 (Python) high-confidence rules distilled from real-codebase audits: (1) committed .bak/.orig backup files, (2) dead abstraction (interface/Protocol + impl + zero external callers), (3) empty-chain seam (interface method whose chain is always empty — silent noop), (4) repetitive decode/apply god function (5+ if key in data { parse; assign } blocks → apply-map or patch DTO), (5) Go-only: god factory file vs composition root (false-positive guard: free-function factory files >500 LoC are legitimate composition roots, split by subsystem not by package), (6) typos in exported API (Exectuor, Recieve, etc.). Each rule has deterministic detection commands, generic examples (User/Order only), structural fixes, and false-positive guards.

v0.4.2

  • Enforce package-prefix convention for all counterfeiter mock filenames and --fake-name values across go-mocking-guide.md, go-patterns.md, go-architecture-patterns.md, go-prometheus-metrics-guide.md, and go-kubernetes-crd-controller-guide.md. Prevents collisions in the flat mocks/ directory when two packages export interfaces with the same name (e.g. formatter.Formatter vs status.Formatter).
  • Update k8s CRD guide snippet with example showing controller-k8s-connector.go / ControllerK8sConnector naming.

v0.4.1

  • Scrub go-kubernetes-crd-controller-guide.md of employer-internal references (sm-octopus, trading examples); replace with public bborbe repo references (bborbe/alert, bborbe/cqrs) and generic example domains
  • Fix invalid Go syntax XPreserveUnknownFields: &true → ptr.To(true) using k8s.io/utils/ptr
  • Rename "What to NOT do" heading to "Antipatterns" per guide style

v0.4.0

  • Add go-kubernetes-crd-controller-guide.md covering CRD types, generated clientset, self-install, event-handler pattern, and deliberate exclusions (no Lister, no WaitForCacheSync, no separate YAML manifest)
  • Add /coding:audit-guide command + guide-auditor agent for auditing coding guides against style/structure/indexing/self-containment standards; explicit forbidden-term grep blocks work-context leakage (seibert, octopus, trading domain, personal paths)
  • Improve go-time-injection.md

v0.3.1

  • Fix agent paths to use canonical plugin path ~/.claude/plugins/marketplaces/coding/docs/
  • Clarify definition-of-done to follow all guides, not just language-specific ones

v0.3.0

  • Rewrite README.md per readme-guide.md (add Overview, Requirements, Quick Start, Contributing; CI and license badges; collapse agents into details; reorder commands; fix license to BSD-2-Clause)
  • Add GitHub Actions CI workflow running make precommit on push and PRs
  • Fix Makefile check-links silent-fail bug (pipe subshell swallowed EXIT=1) and allow directory links (-f → -e); strip anchors before check
  • Add check-json target validating .claude-plugin/plugin.json
  • Add sentinel error naming convention (ErrXxx) and backwards-compat alias pattern to go-error-wrapping-guide.md

v0.2.2

  • commit: Read project CLAUDE.md as step 2 of detection — honors project-specific release checklists (extra files to bump, version-sync rules) that the generic workflow would otherwise miss

v0.2.1

  • Add go-mod dependency fix guide
  • Skip license checks for private repos
  • Improve factory guide
  • Sync plugin version to v0.2.0 and add release checklist to CLAUDE.md
  • Improve /commit to detect unreleased commits since last tag on clean working tree

v0.2.0

  • Add readme-guide.md and claude-md-guide.md for README.md vs CLAUDE.md separation
  • Update documentation-guide.md with CLAUDE.md overview and links to specific guides

v0.1.0

  • Add finder agents for self-contained check-guides
  • Add templates, remove personal paths from agents
  • Add vscode and intellij skills, archive old commands
  • Add go-write-test, improve-guide, go-version commands with agents
  • Remove personal paths from shared plugin docs
  • Add 6 metrics patterns to prometheus guide
  • Split go-quality-assistant into focused agents aligned with docs
  • Add vendor counterfeiter detection to go-test-quality-assistant
  • Add file organization rules to go-quality-assistant

v0.0.3

  • Make plugin self-contained: add all agents referenced by commands
  • Add coding: prefix to agent references in commands
  • Add pre-implementation-assistant, license-assistant, godoc-assistant, go-version-manager, shellcheck-assistant

v0.0.2

  • Trim to 4 essential commands: code-review, pr-review, check-guides, commit
  • Trim to 12 agents required by those commands
  • Remove 9 non-essential commands and 10 unused agents

v0.0.1

  • Restructure as Claude Code plugin with .claude-plugin/ metadata
  • Move all guides into docs/ subdirectory
  • Add 12 shareable slash commands (code-review, pr-review, check-guides, etc.)
  • Add 22 shareable agents (go-quality-assistant, srp-checker, etc.)
  • Add llms.txt index for AI agent discovery
  • Add Makefile with link validation
  • Rewrite README as human-first reference with tables