agentleFS
littlebearappslittlebearapps/pitchdocs/CLAUDE.md

pitchdocs

CLAUDE.md8 starsChanged 8 months agoMITView on GitHub

Generate high-quality public-facing repository documentation with a marketing edge. PitchDocs is a Claude Code plugin (pure Markdown, zero runtime dependencies) with 15 skills, 4 agents (3 pipeline + 1 per-project), 3 auto-loaded rules (1 global + 2 installed), 2 installable rule templates, 16 slash commands (14 active + 2 stubs), 1 opt-in hook, and 24 eval test cases. This repo also has ContextDocs installed, which adds the context-updater agent and context-quality.md rule. This is a 100% Markdown-based plugin — no…

The file

We split this CLAUDE.md into the rules it gives an agent

5 rules. About 0 are common, 0 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Benefit-driven language: Every feature claim traces to actual code.

    line 32
  2. Very specific:

    Adding a skill: Create .claude/skills/<name>/SKILL.md, add a corresponding command in commands/<name>.md, update the features list in …

    line 55
  3. Very specific:

    Adding a command: Create commands/<name>.md with YAML frontmatter, update commands tables in README.md, AGENTS.md, and llms.txt

    line 56
  4. Very specific:

    Adding platform support: Update the platform-profiles skill for new platform equivalents.

    line 59
  5. Very specific:

    Before merging a release PR: Run activation evals from GitHub Actions (Actions → Activation Evals → Run workflow) to confirm skill …

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

PitchDocs

Generate high-quality public-facing repository documentation with a marketing edge. PitchDocs is a Claude Code plugin (pure Markdown, zero runtime dependencies) with 15 skills, 4 agents (3 pipeline + 1 per-project), 3 auto-loaded rules (1 global + 2 installed), 2 installable rule templates, 16 slash commands (14 active + 2 stubs), 1 opt-in hook, and 24 eval test cases. This repo also has ContextDocs installed, which adds the context-updater agent and context-quality.md rule.

Project Architecture

This is a 100% Markdown-based plugin — no JavaScript, no Python, no build step. All knowledge lives in structured YAML+Markdown files:

.claude-plugin/plugin.json      → Plugin manifest (name, version, keywords)
.claude/skills/*/SKILL.md       → 16 reference knowledge modules (loaded on-demand)
.claude/agents/docs-writer.md   → Orchestration agent (coordinates researcher → write → reviewer pipeline)
.claude/agents/docs-researcher.md → Codebase discovery and feature extraction agent
.claude/agents/docs-reviewer.md → Post-generation quality validation agent
.claude/agents/context-updater.md → AI context file updater agent (from ContextDocs, not PitchDocs)
.claude/agents/docs-freshness.md → Installed freshness checker (auto-loaded locally; source: agents/docs-freshness.md)
.claude/rules/content-filter.md → Content filter quick reference (auto-loaded; Claude Code only)
.claude/rules/context-quality.md → AI context file quality standards (from ContextDocs, auto-loaded; Claude Code only)
.claude/rules/doc-standards.md → Installed quality standards (auto-loaded locally; source: rules/doc-standards.md)
.claude/rules/docs-awareness.md → Installed documentation trigger map (auto-loaded locally; source: rules/docs-awareness.md)
rules/doc-standards.md          → Quality standards (installed per-project by /pitchdocs:activate)
rules/docs-awareness.md         → Documentation trigger map (installed per-project by /pitchdocs:activate)
agents/docs-freshness.md        → Freshness checker agent (installed per-project by /pitchdocs:activate)
commands/*.md                   → 16 slash command definitions (14 active + 2 stubs redirecting to ContextDocs)
hooks/*.sh                      → 1 opt-in hook script (Claude Code only, installed by /pitchdocs:activate install strict)

Conventions

  • Australian English: realise, colour, behaviour, licence (noun), license (verb)
  • Conventional Commits: feat:, fix:, docs:, chore: — release-please automates versioning
  • Benefit-driven language: Every feature claim traces to actual code. Pattern: [Feature] so you can [outcome] — [evidence]

Key Files

FilePurpose
plugin.jsonVersion, description, keywords — update on every release
content-filter.mdContent filter quick reference rule — risk levels, fetch commands, chunked writing guidance (auto-loaded globally; Claude Code only)
rules/doc-standards.mdQuality standards template — installed per-project by /pitchdocs:activate. Core standards for tone, benefits, badges
rules/docs-awareness.mdDocumentation trigger map template — installed per-project by /pitchdocs:activate. Suggests PitchDocs commands at documentation moments
agents/docs-freshness.mdFreshness checker agent template — installed per-project by /pitchdocs:activate. Read-only checks with command suggestions
docs-writer.mdOrchestrator agent — lightweight inline research for small projects (< 20 files), full sub-agent research for larger projects, conditional reviewer (skipped for new READMEs), content filter mitigations
docs-researcher.mdCodebase discovery agent — platform detection, feature extraction, security signals, lobby split planning. Only spawned for projects with 20+ files.
docs-reviewer.mdQuality validation agent — checklist, banned phrases scan, GEO scoring, 6-dimension quality rubric. Skipped for new READMEs; runs for updates or with --review.
hooks/*.shContent filter write guard (Claude Code only, installed by /pitchdocs:activate install strict)
upstream-versions.jsonTracks 7 pinned spec versions — checked monthly by GitHub Action
llms.txtAI-readable content index — must be updated when files are added/removed
AGENTS.mdCross-tool AI context (Codex CLI format) — must stay in sync with skills/commands
docs/faq/index.mdProtected — source for marketing-site FAQPage JSON-LD on https://littlebearapps.com/help/pitchdocs/. The site's docs-sync (scripts/docs-sync.config.ts in littlebearapps/littlebearapps.com) hard-fails if this directory is missing. Keep ≥7 question-shaped H2 headings (##); update entries in place — never delete. See Protected Documentation Files in AGENTS.md.
tests/evaluations.json24 command routing test scenarios — used by skill-creator evals

When Modifying This Plugin

  1. Adding a skill: Create .claude/skills/<name>/SKILL.md, add a corresponding command in commands/<name>.md, update the features list in README.md, skills table in AGENTS.md, and llms.txt
  2. Adding a command: Create commands/<name>.md with YAML frontmatter, update commands tables in README.md, AGENTS.md, and llms.txt
  3. Changing quality standards: Edit rules/doc-standards.md — this propagates to all projects that have activated PitchDocs
  4. Updating upstream specs: Edit upstream-versions.json and the corresponding skill content
  5. Adding platform support: Update the platform-profiles skill for new platform equivalents. Existing skills reference it via cross-link.
  6. Bumping version: Handled automatically by release-please from conventional commit messages
  7. Before merging a release PR: Run activation evals from GitHub Actions (Actions → Activation Evals → Run workflow) to confirm skill activation hasn't regressed. Target: 80%+.
  8. Updating docs/faq/index.md: Edit answers in place when content drifts; keep ≥7 question-shaped H2 headings (##). Do not delete or move the file — the marketing-site sync pipeline hard-fails without it. See Protected Documentation Files.

Testing & Validation

PitchDocs includes static validation in CI and supports runtime skill evaluation via the skill-creator plugin.

TestWhat It ChecksHow to Run
Frontmatter validationYAML metadata in skills, commands, agentspython tests/validate-frontmatter.py
Token budgetsSkill/rule/agent file sizes within limitsbash tests/check-token-budgets.sh
llms.txt consistencyAll referenced files exist, no orphansbash tests/validate-llms-txt.sh
Command routing evalsSkills activate for correct promptstests/evaluations.json via skill-creator
Skill quality benchmarksOutput quality with/without pluginskill-creator A/B comparison

Install skill-creator: /plugin marketplace add anthropics/claude-plugins-official then /plugin install skill-creator

Per-Project Activation

PitchDocs commands work globally, but advisory features (quality standards, documentation nudges, freshness checking) are opt-in per-project via /pitchdocs:activate. This prevents noise in private repos that don't need marketing-grade docs.

TierCommandWhat's Installed
None (default)Plugin onlyCommands available, content-filter.md rule auto-loaded
Standard/pitchdocs:activate install+ doc-standards.md rule, docs-awareness.md rule, docs-freshness.md agent
Strict/pitchdocs:activate install strict+ content-filter-guard.sh hook (blocks risky file writes)

Source templates live in rules/, agents/, and hooks/ at the plugin root. The activate command copies them into the project's .claude/ directory.

Relationship to ContextDocs

AI context file management (AGENTS.md, CLAUDE.md, .cursorrules, etc.) has moved to ContextDocs. PitchDocs retains /pitchdocs:ai-context and /pitchdocs:context-guard as stub commands that redirect users to install ContextDocs. The doc-refresh skill delegates Step 5 to ContextDocs if installed. The docs-verify skill uses a lightweight context health check; full scoring requires ContextDocs.

Promotion

Listing and promotion drafts, status tracking, and submission content live in docs/promotion/. This includes awesome list PR content, Reddit post drafts, directory submission templates, and a STATUS.md tracking all listings and their current state. Refer to these files when preparing new submissions or checking on existing ones.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.