Changelog

March 15, 2026 · View on GitHub

All notable changes to this project are documented here.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

2.2.0 (2026-03-15)

Added

  • full plugin review fixes — version sync, hook exit codes, stale references, eval coverage (48dc592)
  • remove out-of-scope skill-authoring skill (44ab591), closes #40

Documentation

  • refresh launch artifacts and bump guide versions for v2.1.0 (#42) (eed52d4)
  • sync AI context files with context-updater agent and context-quality rule (27b5fc8)
  • sync bridge files with corrected agent and rule counts (525c470)

[Unreleased]

Fixed

  • Plugin review issues from v2.1.0 release — Version sync in skill frontmatter, hook exit codes, stale references to moved skills, and evaluation test coverage now properly aligned with production state.

Removed

  • skill-authoring skill — This meta-guide about token budgets for writing Claude Code skills is out of scope for PitchDocs (a documentation plugin). Skill authoring guidance belongs in the plugin-dev plugin instead.

Security

  • GitHub Actions supply chain security — All third-party GitHub Actions are now pinned to specific commit SHAs instead of mutable version tags, preventing malicious updates. CODEOWNERS file added for transparent code review governance.

2.1.0 (2026-03-12)

Added

  • add per-project activation for advisory features (fc57dd2)

Fixed

  • add --plugin-dir to activation eval script for CI compatibility (45866e7)
  • correct last 2 activation eval test expectations (cb3af25)
  • correct test expectations for activation evals (8d3c20a)
  • detect API key auth failures in activation eval pre-flight (1ddc554)
  • detect billing and rate limit errors in activation eval pre-flight (ed6e8cc)
  • install plugin via CLI marketplace for CI activation evals (76156bd)
  • reclassify 2 activation eval tests after failure investigation (c87603d)
  • register plugin in installed_plugins.json for CI activation evals (c371239)
  • use parent directory for --plugin-dir in activation evals (9fe0ded)

Documentation

  • add CI activation eval results and Haiku vs Sonnet comparison (172b371)
  • add headless mode (claude -p) known limitation to changelog and troubleshooting (f407ef1)
  • update roadmap, results, and README for post-v2.0.0 changes (2bbac91)

2.0.0 (2026-03-11)

⚠ BREAKING CHANGES

Added

  • add 3 test artifacts — hook unit tests, banned phrase checker, results doc (dc54272)
  • add 6 CI checks — spell check, actionlint, frontmatter validation, llms.txt consistency, orphan detection, token budgets (e2eb70b)
  • add CI integration — hook tests, banned phrases, activation eval workflow (45372e4)
  • add skill activation eval runner for claude -p testing (2be7a4d)
  • split AI context management into ContextDocs plugin (6320b98)

Known Issues

  • Headless mode skill activation — skills may not auto-trigger via claude -p (non-interactive mode). This is a Claude Code platform limitation affecting project-local plugins, not a PitchDocs bug. Interactive mode works correctly.

Fixed

  • AU English spelling, typos config for brands/HTML, lychee exclude canva.com (8871d20)
  • correct hook file path in llms.txt from gitignored .claude/hooks/ to tracked hooks/ (dd6bce0)
  • fix activation eval runner — stream-json format, remove tool restriction (7c35f1c)
  • heading level increment and dead discussions link (55fcb22)
  • resolve markdown lint errors in RESULTS.md and tutorial (5147950)
  • resolve shellcheck SC2012 and SC2129 in CI workflows (81f6411)
  • revert premature version bump so release-please proposes v2.0.0 (47ad6b5)

Performance

  • slim doc-standards.md auto-loaded rule (~200 token reduction) (6f7f9d2), closes #36
  • split 4 over-budget skills into companion reference files (ed0cc84)

Documentation

  • add activation eval reminder to release workflow (c5cf5dd)
  • add Testing & Validation section to CLAUDE.md (9734907)
  • add tutorial, launch artifacts, ROADMAP, and rewrite docs hub (a222d6c)
  • add Untether integration guide and consolidate cross-references (c1a4d89)
  • add V5 activation eval results (80%) and fix V6 name mismatch (61d75d9)
  • regenerate llms-full.txt for v2.0.0 (c63a87f)
  • update CHANGELOG and README for v1.20.0, add V6 eval results (55e9cce)
  • update CHANGELOG.md with skill activation eval and CI check additions (9ae707b)
  • update guides for v2.0.0 — remove stale context references (7c54cab)
  • update RESULTS.md with Phase 3.5 token budget fix outcomes (8d48cab)

1.19.3 (2026-03-09)

Fixed

  • integrate website notification into release-please workflow (fc7f699)

1.19.2 (2026-03-09)

Fixed

  • resolve Context Guard false positives and add Untether session detection (70ba7c6)

Documentation

  • add website accuracy audit against v1.19.1 (f3d4026)
  • replace demo screenshot TODO with showcase gallery (b7dd753)

1.19.1 (2026-03-09)

Fixed

  • reduce /readme context overhead by 72% for small projects (cd34f85)

Documentation

  • regenerate llms-full.txt with slimmed skills and reference files (a9d3778)

1.19.0 (2026-03-09)

Added

  • reduce auto-loaded context by 41% and fix stale counts across docs (f540de6)

Fixed

  • resolve false-positive broken path warnings in drift check hook (5772356)

Documentation

  • regenerate llms-full.txt with updated counts and slimmed rules (63ef9a1)

1.18.0 (2026-03-09)

Added

  • add dark mode logo support and demo screenshot placeholder (92c4217)
  • extract on-demand skills from auto-loaded rules for 56% context token reduction (#28) (eb0ccb2)

[Unreleased]

Breaking Changes

  • /pitchdocs:ai-context and /pitchdocs:context-guard deprecated — AI context file management is now the responsibility of ContextDocs. These commands remain as stubs that redirect users to install ContextDocs. For projects that had Context Guard hooks enabled, migrate to ContextDocs' hook-based approach for continued drift detection.

Added

  • 6 new automated CI checks — Validate documentation quality in GitHub Actions (or any CI/CD system) so you catch broken links, stale content, and formatting issues before merging. Includes spell checking (typos, misspellings), frontmatter validation (YAML metadata), llms.txt consistency (orphan file detection), banned phrase detection (AI-speak prevention), file orphan detection, and token budget enforcement for skills and rules.

  • Activation evaluation framework — Measure how well PitchDocs commands trigger for the prompts you use. The /pitchdocs:features skill now includes activation eval support via Claude's -p prompt-caching mode, and all 16 skills have been tuned against 20 baseline evaluation scenarios (achieving 80%+ accuracy).

  • Test artifacts and validation results — Track activation eval runs in tests/activation-results/ to catch skill regressions early. New test scripts validate hook content-filter integration, token budgets, frontmatter, and llms.txt consistency.

Documentation

  • Comprehensive getting-started tutorial and workflow guides for new users
  • Launch artifacts generation (Dev.to articles, Hacker News posts, Reddit submissions, promotion drafts)
  • Detailed roadmap with milestone tracking and feature planning visibility
  • Activation evaluation results demonstrating 80% skill routing accuracy across all test scenarios

Fixed

  • Fixed version numbering inconsistency in v2.0.0 and v5 documentation references
  • Resolved false-positive drift check warnings in Context Guard migration period

1.17.0 (2026-03-08)

Added

  • add Signal Gate principle, lifecycle commands, and context health scoring (#25) (8536449)

1.16.0 (2026-03-07)

Added

  • add auto-memory (MEMORY.md) accommodations to AI context guidance (#24) (c68175e)
  • add root-level SKILL.md for directory submissions (7ed7cf9)

Documentation

  • fix stale counts in other-ai-tools compatibility guide (27a7fdb)

1.15.0 (2026-03-06)

Added

  • add two-tier context doc enforcement to Context Guard (46e110f)
  • add user benefits extraction with persona inference and docs-awareness rule (c014db5)

Documentation

  • improve cross-platform documentation for non-Claude Code users (0b6d2a5)
  • sync AI context files with current skill and command counts (30961d2)

1.14.0 (2026-03-05)

Added

  • add GitLab and Bitbucket platform support (3d8fc03)

Documentation

  • strengthen benefit messaging for professional standards, content filter, and context drift (40bfa7c)

1.13.0 (2026-03-05)

Added

  • generate missing AI context files and llms-full.txt (85c2b69)

Documentation

  • trim README to comply with lobby principle and 4-question test (9824332)

1.12.1 (2026-03-04)

Documentation

  • use pitchdocs: namespace prefix for all user-facing slash commands (#18) (2a44f73)

1.12.0 (2026-03-04)

Added

  • add user doc standards — frontmatter, device screenshots, Diátaxis templates (#16) (84f12b1)

1.11.0 (2026-03-04)

Added

  • add workflow, troubleshooting, and reference guides (0a5b82b)

Documentation

  • fix context-quality rule description in AGENTS.md (98b3c77)
  • fix quick start wording and expand documentation links (a425b54)
  • fix stale counts and broken anchor links in guides (11af31a)

1.10.0 (2026-03-02)

Added

  • add content filter mitigation rule, hook, and guidance (#13) (a88db5a)

1.9.0 (2026-03-01)

Added

  • add JTBD mapping, B2B value framework, Time to Hello World, and security credibility extraction (f941dac)
  • add Lobby Principle for README conciseness with structural budgets (8a580ee)
  • recommend emoji+bold+em-dash as default feature list format (65c87ad)

Documentation

  • add paragraph break in What PitchDocs Does section (ec11249)
  • add upstream spec tracking and reusability to features list (fda18eb)
  • clean up README hero and add optional hooks install step (0f25878)
  • condense README body, remove Why section, improve How It Works diagram (e5e11f1)
  • fix mermaid diagram — compact LR layout, use br tags for line breaks (7a8678a)
  • remove By the Numbers table from README (1ebe6e6)
  • rewrite features as emoji+bold+em-dash bullets (Untether style) (1aab3c5)
  • rewrite What PitchDocs Does — problem-first, one paragraph, no diagram (61eb6a7)
  • trim README and move detailed setup guides to docs/ (426bf5a)

1.8.1 (2026-02-28)

Fixed

  • handle both relative and absolute paths in structural change hook (d9562e7)

1.8.0 (2026-02-28)

Added

  • add context-guard hooks for AI context file freshness (ce0e47f)
  • notify littlebearapps.com on new releases (de2ab32)

1.7.0 (2026-02-28)

Added

  • add /doc-refresh command for version-bump documentation updates (58af69b)
  • add AGENTS.md spec to upstream version tracking with v1.1 feature monitoring (c8c9fa3)
  • add cross-platform AI tool setup instructions (3746e05)
  • add cross-platform AI tool setup instructions and AGENTS.md (b0253df)
  • add feature-benefits skill and /features command (d22d9ec)
  • add GEO optimisation, AI context files, docs verification, and launch artifacts (9bd7a5e)
  • add GitHub repository metadata auditing (topics, website, description) (ac16eef)
  • add llms.txt generation, licence guidance, visual assets, and expanded audit (7aa4a91)
  • add npm and PyPI package registry documentation guidance (38540d4)
  • add numeric quality scoring (0-100) to docs-verify with grade bands and CI export (e3a670f)
  • add PitchDocs logo and update README header (1b5f018)
  • add release-please automated versioning (562b0c1)
  • add repo docs, upstream drift detection, and GitHub templates (0a70c43)
  • add security scan check to docs-verify and docs-writer validation (a1987a1)
  • add token budget guidelines to doc-standards and token audit check to docs-verify (053ef50)
  • add version and upstream fields to all skill frontmatter (3989731)
  • add Windsurf, Cline, and Gemini CLI context file generation to ai-context skill (54df3e1)
  • enhance link validation with 4 detection patterns and add docs-ci workflow (2cd88c5)
  • increase README logo size and add logo guidance to skills/rules (c75d350)
  • initial repo-docs plugin with marketing-friendly documentation generation (9b37e73)
  • PitchDocs v1.5.0 — GEO, AI context, docs verification, launch artifacts (2fd32c1)
  • rename plugin from repo-docs to PitchDocs (6843e3f)
  • self-audit improvements — AI context files, comparison table, user guide (6d91796)
  • use project type auto-detection to select writing tone and template in docs-writer agent (7e91f70)

Fixed

  • add content filter mitigations and visual formatting guidance (0e89233)
  • add license embed detection and manifest-match validation to pitchdocs-suite (4d91287)
  • add missing colour to version badge in README (92a679c)
  • improve hero tagline clarity and features section readability (c439044)
  • increase logo breathing room with double br tag (87c7744)
  • reduce logo spacing from double to single br tag (8f4c1d8)
  • refine hero copy with GitHub repo focus, features extraction, and SEO/GEO (8210f31)
  • rework README hero, get started, and use-case sections (64c08d1)
  • use query-param badge URL so release-please preserves colour (f39c1f0)
  • use separate p blocks for README hero spacing (f1ce395)

Documentation

  • add before-after screenshots and move showcase to top of README (6d482d3)
  • add v1.6.0 changelog, bump version to 1.6.0, and update README features (5ad5b75)
  • enhance public repo docs with benefits, comparison, and completeness fixes (2e1cfb3)
  • pre-submission fixes for awesome list and directory submissions (339fa85)
  • propagate enhanced hero, use-case framing, and bold+em-dash patterns into plugin guidance (536412a)
  • remove Australian English enforcement, update README hero (9d962a0)
  • restructure README to follow plugin's own Banesullivan framework (6a09ce0)
  • update getting-started guide and docs hub for v1.6.0 features (ba068dd)
  • update homepage URL to littlebearapps.com/tools/pitchdocs (610f789)
  • update README to reflect new bullets mode and enhanced skill descriptions (1eae311)
  • use Claude logo badge and clarify before/after README examples (dc379bc)

1.6.0 (2026-02-28)

Added

  • Numeric quality scoring (0–100) — /docs-verify score rates documentation across 5 dimensions (completeness, structure, freshness, link health, evidence) with A–F grade bands — CI mode exports PITCHDOCS_SCORE and PITCHDOCS_GRADE, supports --min-score N threshold
  • Security scanning for generated docs — /docs-verify detects leaked credentials, internal paths (/Users/, /home/), and internal hostnames so you can catch accidental exposure before shipping
  • Enhanced link validation — 4 new detection patterns: case-sensitive path checks, fragment-only anchor validation, redirect chain detection, and relative link resolution from nested docs directories
  • Docs CI workflow — ready-to-use .github/workflows/docs-ci.yml with markdownlint-cli2 and lychee link checking, triggered on Markdown changes and monthly schedule
  • Token budget guidelines — skill token cost targets (reference <3K, workflow <4K, combined <5K) in doc-standards rule, plus token audit check in /docs-verify to flag oversized skills
  • Skill version tracking — all 12 skills carry version: and upstream: fields in YAML frontmatter for provenance and drift detection
  • Project type auto-detection — docs-writer agent classifies repos (library, CLI, web-app, API, plugin, docs-site, monorepo) and selects writing tone, hero emphasis, and quick start style automatically
  • Windsurf, Cline, and Gemini CLI context files — /ai-context generates .windsurfrules, .clinerules, and GEMINI.md alongside existing formats (7 AI context files total, 9 AI tools supported)
  • AGENTS.md spec tracking — upstream version monitoring for the AGENTS.md v1.0 spec with v1.1 feature watch (8 upstream specs tracked total)
  • Licence embed detection — pitchdocs-suite validates that verbatim licence text isn't accidentally embedded in skill/rule/context files, and cross-checks manifest license field against the LICENSE file

Changed

  • docs-verify skill version bumped to 1.3.0 (quality scoring, enhanced links, security scan, token audit)
  • ai-context skill version bumped to 1.1.0 (3 new context file formats, spec tracking)
  • doc-standards rule expanded with token budget guidelines section
  • docs-writer agent validation now includes security scan checklist items and project type classification
  • Plugin keywords updated for v1.6.0

Fixed

  • Licence file extension check — flags LICENSE.md (GitHub prefers extensionless LICENSE for automatic detection)

1.5.0 (2026-02-26)

Added

  • GEO (Generative Engine Optimisation) — new section in doc-standards rule with crisp definitions, atomic sections, concrete statistics, comparison tables, TL;DR blocks, and cross-referencing patterns structured for LLM citation
  • GEO patterns in public-readme skill — first-paragraph-as-definition guidance, comparison table optimisation for "X vs Y" queries, and semantic heading hierarchy enforcement
  • Diataxis framework in user-guides skill — classify docs into tutorials, how-to guides, reference, and explanation quadrants with updated directory layout
  • ai-context skill and /ai-context command — generate AGENTS.md, CLAUDE.md, .cursorrules, and .github/copilot-instructions.md from codebase analysis with staleness audit mode
  • docs-verify skill and /docs-verify command — validate broken links, stale content (90-day threshold via git blame), llms.txt sync, heading hierarchy, image alt text, badge URLs, and feature coverage with CI-friendly output
  • launch-artifacts skill and /launch command — transform README/CHANGELOG into Dev.to articles, Hacker News "Show HN" posts, Reddit posts, Twitter/X threads, awesome list submission PRs, and social preview image guidance
  • api-reference skill — configuration templates and comment conventions for TypeDoc, Sphinx/mkdocstrings, godoc, and rustdoc with language auto-detection
  • AI context files in pitchdocs-suite inventory — AGENTS.md and copilot-instructions.md at Tier 2, CLAUDE.md and .cursorrules at Tier 3
  • Diataxis coverage check in /docs-audit — flags missing documentation quadrants
  • AI context staleness check in /docs-audit — verifies context files match current codebase
  • Documentation verification check in /docs-audit — recommends /docs-verify for comprehensive validation
  • Enhanced user guide patterns — copy-paste-ready code examples, error recovery with collapsible troubleshooting, video/screencast placement guidance, and Diataxis cross-links
  • GitHub Actions docs CI template — markdownlint + lychee link checking workflow in docs-verify skill

Changed

  • pitchdocs-suite audit scan now checks for AGENTS.md, CLAUDE.md, .cursorrules, and copilot-instructions.md
  • docs-writer agent now references 4 additional skills (ai-context, docs-verify, launch-artifacts, api-reference)
  • Docs inventory expanded from 17+ to 20+ files across all tiers
  • Plugin version bumped to 1.5.0 with new keywords (seo, geo, ai-context, agents-md, diataxis)

1.4.1 (2026-02-26)

Fixed

  • add content filter mitigations and visual formatting guidance (0e89233)
  • add missing colour to version badge in README (92a679c)
  • use query-param badge URL so release-please preserves colour (f39c1f0)

1.4.0 (2026-02-25)

Added

  • rename plugin from repo-docs to PitchDocs (6843e3f)

1.3.0 (2026-02-25)

Added

  • add npm and PyPI package registry documentation guidance (38540d4)
  • add release-please automated versioning (562b0c1)

1.2.0 - 2026-02-25

Added

  • llms-txt skill — llmstxt.org specification reference with generation patterns for repos and docs sites
  • /llms-txt command — generate llms.txt and llms-full.txt for LLM-friendly content curation
  • LICENSE selection framework with decision guidance in repo-docs-suite skill
  • Visual assets guidance — storage locations, formats, naming conventions, alt text requirements
  • Social preview image audit check (1280×640, Settings reminder)
  • SUPPORT.md template in repo-docs-suite Tier 2
  • .github/release.yml template for auto-generated GitHub Release notes
  • CITATION.cff template (conditional, Tier 3) for academic/research projects
  • llms.txt, SUPPORT.md, release.yml, and CITATION.cff presence checks in /docs-audit
  • Visual assets presence check in /docs-audit
  • GitHub repository metadata checks in /docs-audit — topics, website URL, and description
  • Topic suggestion framework in repo-docs-suite skill based on project type, language, and ecosystem
  • Repository metadata step in docs-writer agent workflow
  • Validation checklist additions in docs-writer agent — visual elements, LICENSE match, social preview, llms.txt

1.1.0 - 2026-02-25

Added

  • feature-benefits skill — systematic codebase scanning for features with evidence-based benefit translation
  • /features command — standalone feature extraction with inventory, table, and audit modes
  • Feature coverage check in /docs-audit — detects undocumented and over-documented features
  • Feature-to-Benefit writing principles in doc-standards rule

Changed

  • docs-writer agent Step 2 now uses the 5-step Feature Extraction Workflow instead of vague bullet points
  • /readme command loads feature-benefits skill and verifies evidence-based features
  • public-readme skill includes guidance on populating features tables from codebase scans

1.0.0 - 2026-02-25

Added

  • /readme command — generate marketing-friendly READMEs with the Daytona/Banesullivan 4-question framework
  • /changelog command — generate changelogs from git history with user-benefit language
  • /roadmap command — generate roadmaps from GitHub milestones and issues
  • /docs-audit command — audit documentation completeness across 14 file types
  • /user-guide command — generate task-oriented user guides in docs/guides/
  • public-readme skill — README structure with hero section, benefit-driven features, and comparison tables
  • changelog skill — Keep a Changelog format with language transformation rules
  • roadmap skill — Roadmap structure from GitHub Projects with emoji status indicators
  • repo-docs-suite skill — complete repo docs inventory with templates for CONTRIBUTING, CODE_OF_CONDUCT, SECURITY, and GitHub templates
  • user-guides skill — task-oriented how-to documentation with hub page and cross-linking
  • docs-writer agent — long-form documentation generation with codebase analysis
  • doc-standards rule — tone, language, badges, and the 4-question framework
  • Monthly upstream spec drift detection via GitHub Actions
  • Upstream version tracking for Keep a Changelog, Contributor Covenant, Conventional Commits, and Semantic Versioning