art-director
FreeOrchestrate iterative visual style searches with branch prompts, decision graphs, feedback loops, and final direction selection.
The English summary is being prepared. Showing the original description.
Use when creating or refactoring CLAUDE.md files - enforces best practices for size, structure, and content organization
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
Creates and refactors CLAUDE.md files following official Anthropic best practices (2025).
| Rule | Why |
|---|---|
| CLAUDE.md < 200 lines | Loads on EVERY request, costs tokens |
| Rules files < 500 lines each | Official recommendation per file |
| Critical rules FIRST | Top = highest priority |
Modular rules → .claude/rules/ | Conditional loading, organized |
Use paths: frontmatter | Load rules only for matching files |
| No linting rules | Use ESLint/Prettier/Biome instead |
| Pointers over copies | Files change, references stay valid |
Claude Code loads memory in this order (higher = higher priority):
| Priority | Type | Location |
|---|---|---|
| Highest | Enterprise | /Library/Application Support/ClaudeCode/CLAUDE.md |
| ↓ | Project | ./CLAUDE.md or ./.claude/CLAUDE.md |
| ↓ | Rules | ./.claude/rules/*.md (conditional) |
| ↓ | User | ~/.claude/CLAUDE.md |
| Lowest | Local | ./CLAUDE.local.md (gitignored) |
Use /memory command to see currently loaded files.
Official recommendation for large projects:
| Tier | Location | Loads | Target |
|---|---|---|---|
| 1. Foundation | CLAUDE.md | Always | < 200 lines |
| 2. Component | .claude/rules/{component}/ | When working in component | < 500 lines |
| 3. Feature | Co-located with code | When working on feature | As needed |
Example structure:
.claude/
├── CLAUDE.md # Tier 1: always loaded
└── rules/
├── database.md # Tier 2: SQL, migrations
├── api.md # Tier 2: API patterns
└── frontend/ # Tier 2: subdirectory
├── components.md # paths: src/**/*.tsx
├── layout.md # paths: src/pages/**/*.tsx
└── tokens.md # paths: **/*.tsx
# Project Name
One-line description.
## Commands
- `npm run dev` - Development
- `npm run build` - Production
- `npm run test` - Tests
## Architecture
| Path | Purpose |
|------|---------|
| `lib/` | Core logic |
| `app/api/` | API routes |
## Key Patterns
**Pattern Name**: One-line explanation.
## Database (if applicable)
| Table | Key Fields |
|-------|------------|
## Modular Docs
See `.claude/rules/` for:
- `database.md` - queries, schema
- `deploy.md` - deployment
## Tech Stack
One line: Next.js 15, PostgreSQL, TypeScript
Use YAML frontmatter for file-type-specific rules:
---
paths: "src/api/**/*.ts"
---
# API Rules
- All endpoints must validate input
- Use standard error format
| Pattern | Matches |
|---|---|
**/*.ts | All .ts files anywhere |
src/**/* | All files under src/ |
*.md | Markdown in project root |
src/components/*.tsx | Components in specific dir |
# Multiple extensions
paths: "src/**/*.{ts,tsx}"
# Multiple directories
paths: "{src,lib}/**/*.ts, tests/**/*.test.ts"
Note: Wrap patterns in quotes for YAML safety.
Rules with paths: only load when working with matching files → saves tokens.
/init for base CLAUDE.md.claude/rules/ for domain-specific docs.claude/rules/:
database.md - queries, schema, connectiondeploy.md - deployment processmessaging.md - integrations (Telegram, etc.)@file references — don't duplicate| Content | Location |
|---|---|
| Project description | CLAUDE.md |
| Critical constraints | CLAUDE.md (top!) |
| Quick start (3 commands) | CLAUDE.md |
| Architecture overview | CLAUDE.md |
| Key patterns (1-liners) | CLAUDE.md |
| SQL queries/schema | .claude/rules/database.md |
| Deployment steps | .claude/rules/deploy.md |
| API documentation | .claude/rules/api.md |
| Git workflow | .claude/rules/git.md |
| Personal preferences | CLAUDE.local.md (gitignored) |
| Code style rules | .eslintrc / biome.json (NOT docs) |
Reference files instead of duplicating:
@README.md
@docs/architecture.md
@~/.claude/snippets/common.md
@docs/file.md@~/path/file.mdPersonal project settings (auto-gitignored):
# My Local Settings
- Prefer verbose output
- Run tests after every change
- My worktree location: .trees/
| Mistake | Fix |
|---|---|
| 500+ lines | Split into .claude/rules/ |
| SQL examples inline | → rules/database.md |
| "Run prettier" rules | Use tool config files |
| Full API docs | → rules/api.md |
| Deployment instructions | → rules/deploy.md |
| Code in CLAUDE.md | Use @file:line references |
| Negative rules only | Add alternatives: "Don't X; use Y instead" |
Before finishing:
.claude/rules/ for domain-specific docs?paths: frontmatter for conditional loading?@ references instead of duplication?| Command | Purpose |
|---|---|
/init | Generate initial CLAUDE.md |
/memory | View loaded memory files |
Official:
Community:
Updated: Jan 2026
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Orchestrate iterative visual style searches with branch prompts, decision graphs, feedback loops, and final direction selection.
The English summary is being prepared. Showing the original description.
Audits the agent rules in the current folder (AGENTS.md, nested AGENTS.md files) and the skill descriptions the agent sees at start, then reports what to cut, move or rewrite and edits only after the user says yes. Use on «аудит правил», «проверь AGENTS.md», «почему агент не слушается правил», «проверь скиллы».
The English summary is being prepared. Showing the original description.
Робот агента Personal Corp. Агент делает себе голову робота из набора деталей, хранит её в штабе в stack/avatars.json и кладёт в штаб страницу avatar/robot.html, где робот стоит за партой своего штаба и вокруг видны отделы, без сервера и интернета. На уроке по ссылке входа с экрана урока робот входит в живой класс и горит маячком, пока агент работает. Триггеры /avatar, /avatar <ссылка входа>, сделай себе аватар, покажи своего робота, приди на урок.
The English summary is being prepared. Showing the original description.
Use when choosing between engines or models for an agent pipeline and the answer must come from measurement on your own data, not from marketing pages: speech recognition for meeting recordings, the model that turns a transcript into notes, the critic model that checks it. Runs every variant in an isolated container, measures time, cost per hour of input (USD and RUB, with price source and date) and quality (WER/CER and course terms for ASR; code checks plus a judge of a different model for LLM steps), writes results.json/csv and one static analytics page. Triggers on "бенчмарк", "сравни движки", "сколько стоит час записи", "какую модель взять для пайплайна", "benchmark the pipeline", "compare ASR engines".
The English summary is being prepared. Showing the original description.
Use when user asks for Claude Code usage stats, weekly analytics, project activity summary, or wants to see what projects were worked on. Triggers on "аналитика", "статистика claude", "cc stats", "weekly report", "что делал"
The English summary is being prepared. Showing the original description.
Use when needing strategic project analysis from multiple independent expert perspectives. Triggers on business decisions, growth strategy, product direction, competitive analysis, or any situation where diverse C-level opinions reduce blind spots
The English summary is being prepared. Showing the original description.