fblissjr/fb-claude-skills

Read-only MCP server over <HOME>/.claude/agent_state.duckdb -- query run history, watermarks, skill versions, and flywheel metrics instead of shelling out to the agent-state CLI.

Auto-generate markdown sections using cogapp

Development conventions with auto-detection: SessionStart hook detects Python/JS projects and injects uv, bun, TDD, version pinning, and doc conventions into context. Skills provide detailed reference on demand including dependency security auditing.

Kimball-style dimensional modeling patterns for DuckDB star schemas in agent systems. SessionStart hook auto-injects principles when DuckDB or fact/dim tables detected.

Deterministic animated explainer films for any domain, on two backends — three.js 3D (cel shading, analytic IK, a per-frame-pure post chain, shots as data with a match-cut constraint) and Canvas2D flat vector — styled by swappable packs and bibles, delivered as self-contained HTML, frame-exact MP4, or animated WebP/AVIF that plays inline in a README

JSON query tool selection and syntax — jg (jsongrep) for fast extraction, jq for transformations

Skills for building and migrating MCP Apps with interactive UIs

MECE decomposition of goals, tasks, and workflows into Agent SDK-ready components with interactive tree visualizer. SessionStart hook auto-injects principles when Agent SDK or decomposition files detected.

Opt a project into down-tier model delegation: installs a standalone .claude/rules/ file telling Claude to route well-specified data and coding tasks to a cheaper model in a subagent, keeping judgment-heavy work in the main loop. Optional pre-shaped .claude/agents/ definitions and an optional agent-state feedback layer.

Enforces a single rule: every path written into a repo must be relative to the repo root. Ships a SessionStart directive plus pre-commit and commit-msg git hooks that hard-block commits whose staged files, message, or branch name reference anything outside the repo.

Tools for analyzing, polishing, and managing Claude Code plugins

Evidence-grounded retrospectives: session and span postmortems of finished work (grounding rule, deviations table, escape analysis, annotate-don't-rewrite) plus a test-suite audit that checks each test's claim, oracle, and reachability envelope, with keep/rewrite/delete verdicts

SessionStart hook that drops a personal, locally-git-ignored pyrightconfig.json into any Python project lacking a Pyright config -- points Pyright at the uv .venv so imports resolve and silences the reportMissingImports flood. Self-healing (adds the venv pointer once .venv appears), never touches a config it didn't write byte-for-byte, no-op outside Python projects.

Search, save, and surface your Readwise Reader library via MCP server with OAuth, DuckDB storage, and full-text search

Pre-share scanner: wraps simonw/scan-for-secrets for literal matches (with escape variants) and composes a ripgrep regex pass for privacy-sensitive paths, emails, API key shapes, and SSH fingerprints.

Deterministic animated films of any register — explainer, game cutscene, meme, character short — on the three.js node stack (WebGPURenderer with WebGL2 fallback, TSL node materials, MaterialX procedural noise) plus a Canvas2D backend, delivered as self-contained HTML, frame-exact MP4, or animated WebP/AVIF. Successor to explainer-video: same beats/shots/instruments method, new renderer and material ceiling.

Interactive MCP App dashboard: quality checks, token budgets, freshness, and repo hygiene

Maintenance tools for skills repos: validation, quality, freshness, upstream detection, best practices review, wiki-sanity lint

Design functional, readable terminal UIs with proper visual hierarchy, semantic color, and responsive layout. SessionStart hook auto-injects principles when Rich/Textual/Click detected.

Writing skills for clear, accessible prose — an American plain-language house style (plain English, active voice, front-loaded content, sentence case, no bold for emphasis) plus voice matching that learns the user's own style.