[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M76SJS129GD96MNZGVS0NSAZ":3,"skill-same:01M76SJS129GD96MNZGVS0NSAZ":73},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M76SJS129GD96MNZGVS0NSAZ","claude-memory",null,"Create and optimize CLAUDE.md memory files or .claude\u002Frules\u002F modular rules for Claude Code projects. Comprehensive guidance on file hierarchy, content structure, path-scoped rules, best practices, and anti-patterns. Use when working with CLAUDE.md files, .claude\u002Frules directories, setting up new projects, or improving Claude Code's context awareness.","zuzu59\u002Fz-skills","zuzu59","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F15345233?v=4","01M756RFXRYFCM8DCAHTP0F5YW","Christian Zufferey","01M756RFYZRYH7F070CKQRMW38",0,"MIT","https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fclaude-memory\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: claude-memory\ndescription: Create and optimize CLAUDE.md memory files or .claude\u002Frules\u002F modular rules for Claude Code projects. Comprehensive guidance on file hierarchy, content structure, path-scoped rules, best practices, and anti-patterns. Use when working with CLAUDE.md files, .claude\u002Frules directories, setting up new projects, or improving Claude Code's context awareness.\nargument-hint: [init | optimize | task description]\n---\n\n\u003Ccore_principle>\nMemory files consume tokens from the context window. ~100-150 instruction slots available for your customizations. Keep files minimal — only include what the agent cannot discover on its own.\n\n**Two approaches:**\n- **CLAUDE.md** — Single file, best for small projects (\u003C 100 lines)\n- **.claude\u002Frules\u002F** — Modular files with optional path-scoping, best for large projects\n\u003C\u002Fcore_principle>\n\n\u003Cquick_start>\nRun `\u002Finit` to auto-generate a CLAUDE.md. Or create manually:\n\n```markdown\n# Project Name\n\n## Tech Stack\n- [Primary framework]\n- [Key non-obvious libraries]\n\n## Commands\n- `npm run dev` - Dev server\n- `npm test` - Run tests\n- `npm run build` - Build\n\n## Rules\n- [2-3 critical project-specific rules]\n```\n\n- Press `#` during a session to add memory items quickly\n- Use `\u002Fmemory` to open CLAUDE.md in your editor\n\u003C\u002Fquick_start>\n\n\u003Cfile_hierarchy>\n\n| Priority | Location | Scope |\n|----------|----------|-------|\n| 1 (Highest) | Enterprise policy (managed by IT) | All org users |\n| 2 | `.\u002FCLAUDE.md` or `.\u002F.claude\u002FCLAUDE.md` | Team via git |\n| 2 | `.\u002F.claude\u002Frules\u002F*.md` | Team via git |\n| 3 | `~\u002F.claude\u002FCLAUDE.md` | All your projects |\n| 3 | `~\u002F.claude\u002Frules\u002F*.md` | All your projects |\n| 4 (Lowest) | `.\u002FCLAUDE.local.md` (auto-gitignored) | Just you |\n\nClaude recurses UP from current directory, loading all CLAUDE.md files found. Also discovers CLAUDE.md in subtrees when reading files in those directories.\n\n**Monorepo strategy:** Root file defines WHEN; subtree files define HOW.\n```\nroot\u002FCLAUDE.md           # Universal: tech stack, git workflow\napps\u002Fweb\u002FCLAUDE.md       # Frontend-specific\napps\u002Fapi\u002FCLAUDE.md       # Backend-specific\n```\n\u003C\u002Ffile_hierarchy>\n\n\u003Crules_directory>\nThe `.claude\u002Frules\u002F` directory splits instructions into focused markdown files.\n\n**Use `.claude\u002Frules\u002F` when:** many concerns, different rules for different file types, team maintains different areas.\n**Use CLAUDE.md when:** small project, universal rules, single source of truth.\n\nPath-scoped rules use YAML frontmatter:\n```yaml\n---\npaths:\n  - \"src\u002Fapi\u002F**\u002F*.ts\"\n---\n# API Rules\n- All endpoints must include input validation\n```\n\nSupported patterns: `**\u002F*.ts`, `src\u002F**\u002F*`, `src\u002F**\u002F*.{ts,tsx}`, `{src,lib}\u002F**\u002F*.ts`\n\nRules without `paths` frontmatter load unconditionally.\n\nSee [references\u002Frules-directory-guide.md](references\u002Frules-directory-guide.md) for complete guide including symlinks, user-level rules, and migration.\n\u003C\u002Frules_directory>\n\n\u003Ccontent_structure>\nStructure CLAUDE.md with only these sections:\n\n1. **Project purpose** (1-3 lines) — What the project is\n2. **Tech stack** (compact) — Only non-obvious technologies\n3. **Commands** — Non-obvious dev, build, and test commands\n4. **Important files** — Architecture-critical, non-obvious files only\n5. **Rules** — Prohibitions and constraints that prevent mistakes (highest-value lines)\n6. **Workflow** (optional) — Only if non-standard\n\n**Do NOT include:**\n- Repository overviews (agent discovers structure itself)\n- Code style rules (linters enforce these)\n- Generic best practices (\"write clean code\", \"DRY\", \"SOLID\")\n- Redundant specs (copies of config files, env vars, schema descriptions)\n- Marketing\u002Fgoals (vision statements, KPIs, roadmaps)\n- Verbose explanations (paragraphs where one line suffices)\n\nSee [references\u002Fsection-templates.md](references\u002Fsection-templates.md) for ready-to-use templates.\nSee [references\u002Fproject-patterns.md](references\u002Fproject-patterns.md) for framework-specific patterns.\n\u003C\u002Fcontent_structure>\n\n\u003Cwriting_rules>\n**Golden rule:** If someone with zero project context reads your CLAUDE.md and gets confused, Claude will too.\n\n**Be specific, never vague:**\n```\n❌ \"Format code properly\" \u002F \"Write good tests\" \u002F \"Follow best practices\"\n✅ \"Run `pnpm lint` before committing\" \u002F \"Tests in `__tests__\u002F` using Vitest\"\n```\n\n**Prohibitions > positive guidance:**\n```\n❌ \"Try to use TanStack Form for forms\"\n✅ \"NEVER use native form\u002FuseState for forms — ALWAYS use TanStack Form\"\n```\n\n**Show, don't tell:** When format matters, show a concrete example (3-5 lines max).\n\n**Emphasis hierarchy:** CRITICAL > NEVER > ALWAYS > IMPORTANT > YOU MUST\n- Put critical rules **first** in each section\n- Use **bold + keyword** for non-negotiable rules: `**CRITICAL**: Never commit secrets`\n\nSee [references\u002Fprompting-techniques.md](references\u002Fprompting-techniques.md) for advanced techniques.\n\u003C\u002Fwriting_rules>\n\n\u003Csize_limits>\n- **Ideal:** \u003C 100 lines\n- **Maximum:** 150 lines before performance degrades\n- **Over 200 lines:** directives start getting lost\n\nWhen exceeding limits, split into `.claude\u002Frules\u002F` files or link to separate docs:\n```markdown\n- **API patterns**: See [docs\u002Fapi-patterns.md](docs\u002Fapi-patterns.md)\n- **Testing guide**: See [docs\u002Ftesting-guide.md](docs\u002Ftesting-guide.md)\n```\n\nCLAUDE.md supports importing: `@docs\u002Fcoding-standards.md` (relative\u002Fabsolute paths, `~` expansion, up to 5 levels deep, not evaluated inside code blocks).\n\u003C\u002Fsize_limits>\n\n\u003Cworkflow>\n**ALWAYS ASK FIRST: Storage Strategy**\n\nBefore creating or updating memory files, use AskUserQuestion:\n- **Option 1: Single CLAUDE.md** — \u003C 100 lines, simple project, universal rules\n- **Option 2: Modular .claude\u002Frules\u002F** — 100+ lines, different rules for different files\n\n**Creating new memory:**\n1. Start with `\u002Finit` or minimal template\n2. Add tech stack and commands first\n3. Add rules only as you encounter friction\n4. Test with real tasks, iterate based on Claude's behavior\n\n**Maintaining:**\n1. Review quarterly or when project changes significantly\n2. Remove outdated instructions\n3. Add patterns that required repeated explanation\n4. Use `#` for quick additions during work\n\n**Troubleshooting:**\n\n| Problem | Solution |\n|---------|----------|\n| Claude ignores instructions | Reduce file size, add emphasis (CRITICAL, NEVER) |\n| Context overflow | Use `\u002Fclear`, split into .claude\u002Frules\u002F |\n| Instructions conflict | Consolidate, use hierarchy (root vs subtree) |\n| Path rules not applying | Verify glob pattern matches target files |\n\u003C\u002Fworkflow>\n\n\u003Cinit_workflow>\n## `\u002Fclaude-memory init` — Create Minimal CLAUDE.md\n\nWhen the user argument contains \"init\", execute this workflow to scaffold a CLAUDE.md with only the essential sections.\n\n### Step 1: Detect project context\n\nRead `package.json` (or equivalent: `pyproject.toml`, `Cargo.toml`, `go.mod`, etc.) to detect:\n- Project name\n- Tech stack (framework, language)\n- Available scripts\u002Fcommands (dev, build, test)\n\nAlso check for existing linter configs (ESLint, Biome, Prettier, tsconfig) — do NOT include rules they already enforce.\n\n### Step 2: Identify important files\n\nScan for non-obvious architecture-critical files. Look for:\n- Auth config files\n- Server action wrappers, API helpers\n- Database schema files\n- Custom middleware or shared utilities\n- Skip standard framework files (package.json, tsconfig, next.config, etc.)\n\n### Step 3: Generate CLAUDE.md\n\nCreate the file with ONLY these sections. Each section should be compact — the whole file should be under 50 lines.\n\n```markdown\n# [Project Name]\n[One-line description of what this project is]\n\n## Tech Stack\n- [Only non-obvious technologies — skip deps detectable from config files]\n\n## Commands\n- `[dev command]` - Dev server\n- `[build command]` - Build\n- `[test command]` - Tests\n- [Any non-obvious commands like db:push, seed, etc.]\n\n## Important Files\n- [Only files the agent wouldn't discover naturally]\n\n## Rules\n- [Leave empty with a comment: \"Add project-specific rules as you encounter friction\"]\n```\n\n### Step 4: Present and write\n\nShow the generated file to the user. Write it to `.\u002FCLAUDE.md` after approval.\n\n**Rules:**\n- NEVER include directory structure, code style rules, generic best practices, or marketing text\n- NEVER include anything a linter or TypeScript already enforces\n- Target: 20-50 lines. If it's longer, you're including too much\n- The Rules section starts empty — it gets populated over time as mistakes happen\n\u003C\u002Finit_workflow>\n\n\u003Coptimize_workflow>\n## `\u002Fclaude-memory optimize` — Deep CLAUDE.md Cleanup\n\nWhen the user argument contains \"optimize\", execute this workflow.\n\n**CRITICAL — Step 0 is MANDATORY. Do NOT skip it. Do NOT start optimizing without reading the guide first.**\n\n### Step 0: Read the optimization guide (REQUIRED FIRST)\n\nYOU MUST use the Read tool on `{SKILL_PATH}\u002Freferences\u002Foptimize-guide.md` BEFORE doing anything else.\nThis file contains the research data (ETH Zurich study), the 6 bloat categories with specific examples, target metrics, and before\u002Fafter examples. Without reading it, you will miss removal criteria and produce a subpar optimization.\n\n### Step 1: Inventory\n\nRead every CLAUDE.md, CLAUDE.local.md, and `.claude\u002Frules\u002F*.md` in the project. Count total lines.\n\n### Step 2: Read linter configs\n\nRead ESLint\u002FBiome\u002FPrettier\u002FTypeScript configs. Any CLAUDE.md line duplicating an enforced rule → delete.\n\n### Step 3: Apply the 6 bloat categories from the guide\n\nFor each line ask: \"Can the agent discover this by reading the project, or does a linter enforce this?\" If yes → delete.\n\nRemove everything matching:\n1. Linter-enforced rules (ESLint, Prettier, Biome, TypeScript strict)\n2. Marketing \u002F goals \u002F vision (zero code value)\n3. Obvious info the agent discovers itself (directory structure, framework defaults, deps from package.json)\n4. Verbose explanations (paragraphs where 1 line suffices, tutorials, history)\n5. Redundant specs (copies of config files, schema descriptions, env var lists)\n6. Generic best practices (\"write clean code\", \"DRY\", \"SOLID\")\n\n### Step 4: Keep only essentials\n\n- Project purpose (1-3 lines)\n- Tech stack (compact, non-obvious only)\n- Core commands (non-obvious only)\n- Testing commands\n- Important files (non-obvious only)\n- Project-specific rules (prohibitions + constraints)\n- Workflow (only if non-standard)\n\n### Step 5: Compress\n\n- Paragraphs → bullet points\n- 3-line rules → 1-line rules\n- Zero filler words (\"In order to\", \"It's important to note that\")\n- Merge related items\n\n### Step 6: Present diff\n\nShow before\u002Fafter with line counts. For each removal, cite which bloat category it falls under.\nLet user approve before applying changes.\n\n**Target:** \u003C 100 lines ideal, \u003C 150 max.\n\u003C\u002Foptimize_workflow>\n\n\u003Creference_guides>\n- **Optimization guide**: [references\u002Foptimize-guide.md](references\u002Foptimize-guide.md) — Research-backed bloat checklist, 6 removal categories, before\u002Fafter examples\n- **Rules directory**: [references\u002Frules-directory-guide.md](references\u002Frules-directory-guide.md) — Complete .claude\u002Frules\u002F guide with path-scoping, YAML syntax, symlinks, migration\n- **Prompting techniques**: [references\u002Fprompting-techniques.md](references\u002Fprompting-techniques.md) — Emphasis strategies, clarity techniques, constraint patterns\n- **Section templates**: [references\u002Fsection-templates.md](references\u002Fsection-templates.md) — Copy-paste templates for each section type\n- **Comprehensive example**: [references\u002Fcomprehensive-example.md](references\u002Fcomprehensive-example.md) — Full production SaaS CLAUDE.md\n- **Project patterns**: [references\u002Fproject-patterns.md](references\u002Fproject-patterns.md) — Next.js, Express, Python, Monorepo patterns\n\u003C\u002Freference_guides>\n","\u003Ccore_principle>\nMemory files consume tokens from the context window. ~100-150 instruction slots available for your customizations. Keep files minimal — only include what the agent cannot discover on its own.\n\n**Two approaches:**\n- **CLAUDE.md** — Single file, best for small projects (\u003C 100 lines)\n- **.claude\u002Frules\u002F** — Modular files with optional path-scoping, best for large projects\n\u003C\u002Fcore_principle>\n\n\u003Cquick_start>\nRun `\u002Finit` to auto-generate a CLAUDE.md. Or create manually:\n\n```markdown\n# Project Name\n\n## Tech Stack\n- [Primary framework]\n- [Key non-obvious libraries]\n\n## Commands\n- `npm run dev` - Dev server\n- `npm test` - Run tests\n- `npm run build` - Build\n\n## Rules\n- [2-3 critical project-specific rules]\n```\n\n- Press `#` during a session to add memory items quickly\n- Use `\u002Fmemory` to open CLAUDE.md in your editor\n\u003C\u002Fquick_start>\n\n\u003Cfile_hierarchy>\n\n| Priority | Location | Scope |\n|----------|----------|-------|\n| 1 (Highest) | Enterprise policy (managed by IT) | All org users |\n| 2 | `.\u002FCLAUDE.md` or `.\u002F.claude\u002FCLAUDE.md` | Team via git |\n| 2 | `.\u002F.claude\u002Frules\u002F*.md` | Team via git |\n| 3 | `~\u002F.claude\u002FCLAUDE.md` | All your projects |\n| 3 | `~\u002F.claude\u002Frules\u002F*.md` | All your projects |\n| 4 (Lowest) | `.\u002FCLAUDE.local.md` (auto-gitignored) | Just you |\n\nClaude recurses UP from current directory, loading all CLAUDE.md files found. Also discovers CLAUDE.md in subtrees when reading files in those directories.\n\n**Monorepo strategy:** Root file defines WHEN; subtree files define HOW.\n```\nroot\u002FCLAUDE.md           # Universal: tech stack, git workflow\napps\u002Fweb\u002FCLAUDE.md       # Frontend-specific\napps\u002Fapi\u002FCLAUDE.md       # Backend-specific\n```\n\u003C\u002Ffile_hierarchy>\n\n\u003Crules_directory>\nThe `.claude\u002Frules\u002F` directory splits instructions into focused markdown files.\n\n**Use `.claude\u002Frules\u002F` when:** many concerns, different rules for different file types, team maintains different areas.\n**Use CLAUDE.md when:** small project, universal rules, single source of truth.\n\nPath-scoped rules use YAML frontmatter:\n```yaml\n---\npaths:\n  - \"src\u002Fapi\u002F**\u002F*.ts\"\n---\n# API Rules\n- All endpoints must include input validation\n```\n\nSupported patterns: `**\u002F*.ts`, `src\u002F**\u002F*`, `src\u002F**\u002F*.{ts,tsx}`, `{src,lib}\u002F**\u002F*.ts`\n\nRules without `paths` frontmatter load unconditionally.\n\nSee [references\u002Frules-directory-guide.md](references\u002Frules-directory-guide.md) for complete guide including symlinks, user-level rules, and migration.\n\u003C\u002Frules_directory>\n\n\u003Ccontent_structure>\nStructure CLAUDE.md with only these sections:\n\n1. **Project purpose** (1-3 lines) — What the project is\n2. **Tech stack** (compact) — Only non-obvious technologies\n3. **Commands** — Non-obvious dev, build, and test commands\n4. **Important files** — Architecture-critical, non-obvious files only\n5. **Rules** — Prohibitions and constraints that prevent mistakes (highest-value lines)\n6. **Workflow** (optional) — Only if non-standard\n\n**Do NOT include:**\n- Repository overviews (agent discovers structure itself)\n- Code style rules (linters enforce these)\n- Generic best practices (\"write clean code\", \"DRY\", \"SOLID\")\n- Redundant specs (copies of config files, env vars, schema descriptions)\n- Marketing\u002Fgoals (vision statements, KPIs, roadmaps)\n- Verbose explanations (paragraphs where one line suffices)\n\nSee [references\u002Fsection-templates.md](references\u002Fsection-templates.md) for ready-to-use templates.\nSee [references\u002Fproject-patterns.md](references\u002Fproject-patterns.md) for framework-specific patterns.\n\u003C\u002Fcontent_structure>\n\n\u003Cwriting_rules>\n**Golden rule:** If someone with zero project context reads your CLAUDE.md and gets confused, Claude will too.\n\n**Be specific, never vague:**\n```\n❌ \"Format code properly\" \u002F \"Write good tests\" \u002F \"Follow best practices\"\n✅ \"Run `pnpm lint` before committing\" \u002F \"Tests in `__tests__\u002F` using Vitest\"\n```\n\n**Prohibitions > positive guidance:**\n```\n❌ \"Try to use TanStack Form for forms\"\n✅ \"NEVER use native form\u002FuseState for forms — ALWAYS use TanStack Form\"\n```\n\n**Show, don't tell:** When format matters, show a concrete example (3-5 lines max).\n\n**Emphasis hierarchy:** CRITICAL > NEVER > ALWAYS > IMPORTANT > YOU MUST\n- Put critical rules **first** in each section\n- Use **bold + keyword** for non-negotiable rules: `**CRITICAL**: Never commit secrets`\n\nSee [references\u002Fprompting-techniques.md](references\u002Fprompting-techniques.md) for advanced techniques.\n\u003C\u002Fwriting_rules>\n\n\u003Csize_limits>\n- **Ideal:** \u003C 100 lines\n- **Maximum:** 150 lines before performance degrades\n- **Over 200 lines:** directives start getting lost\n\nWhen exceeding limits, split into `.claude\u002Frules\u002F` files or link to separate docs:\n```markdown\n- **API patterns**: See [docs\u002Fapi-patterns.md](docs\u002Fapi-patterns.md)\n- **Testing guide**: See [docs\u002Ftesting-guide.md](docs\u002Ftesting-guide.md)\n```\n\nCLAUDE.md supports importing: `@docs\u002Fcoding-standards.md` (relative\u002Fabsolute paths, `~` expansion, up to 5 levels deep, not evaluated inside code blocks).\n\u003C\u002Fsize_limits>\n\n\u003Cworkflow>\n**ALWAYS ASK FIRST: Storage Strategy**\n\nBefore creating or updating memory files, use AskUserQuestion:\n- **Option 1: Single CLAUDE.md** — \u003C 100 lines, simple project, universal rules\n- **Option 2: Modular .claude\u002Frules\u002F** — 100+ lines, different rules for different files\n\n**Creating new memory:**\n1. Start with `\u002Finit` or minimal template\n2. Add tech stack and commands first\n3. Add rules only as you encounter friction\n4. Test with real tasks, iterate based on Claude's behavior\n\n**Maintaining:**\n1. Review quarterly or when project changes significantly\n2. Remove outdated instructions\n3. Add patterns that required repeated explanation\n4. Use `#` for quick additions during work\n\n**Troubleshooting:**\n\n| Problem | Solution |\n|---------|----------|\n| Claude ignores instructions | Reduce file size, add emphasis (CRITICAL, NEVER) |\n| Context overflow | Use `\u002Fclear`, split into .claude\u002Frules\u002F |\n| Instructions conflict | Consolidate, use hierarchy (root vs subtree) |\n| Path rules not applying | Verify glob pattern matches target files |\n\u003C\u002Fworkflow>\n\n\u003Cinit_workflow>\n## `\u002Fclaude-memory init` — Create Minimal CLAUDE.md\n\nWhen the user argument contains \"init\", execute this workflow to scaffold a CLAUDE.md with only the essential sections.\n\n### Step 1: Detect project context\n\nRead `package.json` (or equivalent: `pyproject.toml`, `Cargo.toml`, `go.mod`, etc.) to detect:\n- Project name\n- Tech stack (framework, language)\n- Available scripts\u002Fcommands (dev, build, test)\n\nAlso check for existing linter configs (ESLint, Biome, Prettier, tsconfig) — do NOT include rules they already enforce.\n\n### Step 2: Identify important files\n\nScan for non-obvious architecture-critical files. Look for:\n- Auth config files\n- Server action wrappers, API helpers\n- Database schema files\n- Custom middleware or shared utilities\n- Skip standard framework files (package.json, tsconfig, next.config, etc.)\n\n### Step 3: Generate CLAUDE.md\n\nCreate the file with ONLY these sections. Each section should be compact — the whole file should be under 50 lines.\n\n```markdown\n# [Project Name]\n[One-line description of what this project is]\n\n## Tech Stack\n- [Only non-obvious technologies — skip deps detectable from config files]\n\n## Commands\n- `[dev command]` - Dev server\n- `[build command]` - Build\n- `[test command]` - Tests\n- [Any non-obvious commands like db:push, seed, etc.]\n\n## Important Files\n- [Only files the agent wouldn't discover naturally]\n\n## Rules\n- [Leave empty with a comment: \"Add project-specific rules as you encounter friction\"]\n```\n\n### Step 4: Present and write\n\nShow the generated file to the user. Write it to `.\u002FCLAUDE.md` after approval.\n\n**Rules:**\n- NEVER include directory structure, code style rules, generic best practices, or marketing text\n- NEVER include anything a linter or TypeScript already enforces\n- Target: 20-50 lines. If it's longer, you're including too much\n- The Rules section starts empty — it gets populated over time as mistakes happen\n\u003C\u002Finit_workflow>\n\n\u003Coptimize_workflow>\n## `\u002Fclaude-memory optimize` — Deep CLAUDE.md Cleanup\n\nWhen the user argument contains \"optimize\", execute this workflow.\n\n**CRITICAL — Step 0 is MANDATORY. Do NOT skip it. Do NOT start optimizing without reading the guide first.**\n\n### Step 0: Read the optimization guide (REQUIRED FIRST)\n\nYOU MUST use the Read tool on `{SKILL_PATH}\u002Freferences\u002Foptimize-guide.md` BEFORE doing anything else.\nThis file contains the research data (ETH Zurich study), the 6 bloat categories with specific examples, target metrics, and before\u002Fafter examples. Without reading it, you will miss removal criteria and produce a subpar optimization.\n\n### Step 1: Inventory\n\nRead every CLAUDE.md, CLAUDE.local.md, and `.claude\u002Frules\u002F*.md` in the project. Count total lines.\n\n### Step 2: Read linter configs\n\nRead ESLint\u002FBiome\u002FPrettier\u002FTypeScript configs. Any CLAUDE.md line duplicating an enforced rule → delete.\n\n### Step 3: Apply the 6 bloat categories from the guide\n\nFor each line ask: \"Can the agent discover this by reading the project, or does a linter enforce this?\" If yes → delete.\n\nRemove everything matching:\n1. Linter-enforced rules (ESLint, Prettier, Biome, TypeScript strict)\n2. Marketing \u002F goals \u002F vision (zero code value)\n3. Obvious info the agent discovers itself (directory structure, framework defaults, deps from package.json)\n4. Verbose explanations (paragraphs where 1 line suffices, tutorials, history)\n5. Redundant specs (copies of config files, schema descriptions, env var lists)\n6. Generic best practices (\"write clean code\", \"DRY\", \"SOLID\")\n\n### Step 4: Keep only essentials\n\n- Project purpose (1-3 lines)\n- Tech stack (compact, non-obvious only)\n- Core commands (non-obvious only)\n- Testing commands\n- Important files (non-obvious only)\n- Project-specific rules (prohibitions + constraints)\n- Workflow (only if non-standard)\n\n### Step 5: Compress\n\n- Paragraphs → bullet points\n- 3-line rules → 1-line rules\n- Zero filler words (\"In order to\", \"It's important to note that\")\n- Merge related items\n\n### Step 6: Present diff\n\nShow before\u002Fafter with line counts. For each removal, cite which bloat category it falls under.\nLet user approve before applying changes.\n\n**Target:** \u003C 100 lines ideal, \u003C 150 max.\n\u003C\u002Foptimize_workflow>\n\n\u003Creference_guides>\n- **Optimization guide**: [references\u002Foptimize-guide.md](references\u002Foptimize-guide.md) — Research-backed bloat checklist, 6 removal categories, before\u002Fafter examples\n- **Rules directory**: [references\u002Frules-directory-guide.md](references\u002Frules-directory-guide.md) — Complete .claude\u002Frules\u002F guide with path-scoping, YAML syntax, symlinks, migration\n- **Prompting techniques**: [references\u002Fprompting-techniques.md](references\u002Fprompting-techniques.md) — Emphasis strategies, clarity techniques, constraint patterns\n- **Section templates**: [references\u002Fsection-templates.md](references\u002Fsection-templates.md) — Copy-paste templates for each section type\n- **Comprehensive example**: [references\u002Fcomprehensive-example.md](references\u002Fcomprehensive-example.md) — Full production SaaS CLAUDE.md\n- **Project patterns**: [references\u002Fproject-patterns.md](references\u002Fproject-patterns.md) — Next.js, Express, Python, Monorepo patterns\n\u003C\u002Freference_guides>\n",{"name":5,"description":7,"argument-hint":24},[25],"init | optimize | task description","master","46a82833a3f7f6f5f408511246a55e0485c5f7ab","skills\u002Fclaude-memory",1,"https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills","https:\u002F\u002Fgithub.com\u002Fzuzu59","2026-09-30 17:58:42",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add zuzu59\u002Fz-skills --skill claude-memory",{"tool":39,"command":40},"bunx","bunx skills add zuzu59\u002Fz-skills --skill claude-memory",{"tool":42,"command":43},"pnpm","pnpm dlx skills add zuzu59\u002Fz-skills --skill claude-memory","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fclaude-memory\u002FSKILL.md",true,[47,53,57,61,65,69],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Fcomprehensive-example.md","REFERENCE","参考文档",5368,"https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fclaude-memory\u002Freferences\u002Fcomprehensive-example.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"references\u002Foptimize-guide.md",10608,"https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fclaude-memory\u002Freferences\u002Foptimize-guide.md",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Fproject-patterns.md",7202,"https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fclaude-memory\u002Freferences\u002Fproject-patterns.md",{"path":62,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":63,"sourceHtmlUrl":64},"references\u002Fprompting-techniques.md",9620,"https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fclaude-memory\u002Freferences\u002Fprompting-techniques.md",{"path":66,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Frules-directory-guide.md",7786,"https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fclaude-memory\u002Freferences\u002Frules-directory-guide.md",{"path":70,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fsection-templates.md",6681,"https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fclaude-memory\u002Freferences\u002Fsection-templates.md",[74,81,88,95,102,109],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M76SQGVSKRYNQX4AA4RXV3K0","writing-great-skills","Reference for writing and editing skills well — the vocabulary and principles that make a skill predictable.","https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fwriting-great-skills\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M76SQDSAZSA1272D56JY8Q2G","use-style","Apply named visual style guides to landing pages, app shells, and UI. Use for $use-style, \u002Fuseskill, list styles, or styles like ios-app, grid, vercel, black-grid, stripe, linear, raycast, gumroad, dusk, or luma.","https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fuse-style\u002FSKILL.md",[],[],{"skillNo":89,"name":90,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":92,"categories":93,"keywords":94,"publishedAt":19},"01M76SQAXDVX02V4JH4PF1MC9G","use-goal","Use when the user asks to create, draft, set, start, or refine a Codex or Claude Code `\u002Fgoal` objective for persistent multi-turn work.","https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fuse-goal\u002FSKILL.md",[],[],{"skillNo":96,"name":97,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":101,"publishedAt":19},"01M76SQ8B11GTTGQ5CZHHGXFXM","use-fable","Orchestrator mode for Fable: Fable thinks, plans, and reviews while Codex CLI and subagents execute all heavy work. Use when the user invokes \u002Fuse-fable, says 'use fable', 'fable mode', 'orchestrator mode', or wants to save Fable tokens\u002Frate limits.","https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fuse-fable\u002FSKILL.md",[],[],{"skillNo":103,"name":104,"nameZh":6,"description":105,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":106,"categories":107,"keywords":108,"publishedAt":19},"01M76SQ4SB7Q4CXEGGPA75KGFK","use-delegate","Delegation mode: the host agent (Claude or Codex) plans and reviews while heavy work runs on cheap executors: OpenCode Kimi K3, Codex GPT-5.6 terra\u002Fsol. Use when the user invokes \u002Fuse-delegate, says 'use delegate', 'delegate mode', 'orchestrator mode', or wants to save tokens\u002Frate limits.","https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fuse-delegate\u002FSKILL.md",[],[],{"skillNo":110,"name":111,"nameZh":6,"description":112,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":113,"categories":114,"keywords":115,"publishedAt":19},"01M76SQ24836GDGMGGPV6Z2B7N","use-artifacts","Create Claude-style local HTML artifacts under ~\u002F.agents\u002Fartifacts. Use for plans, prototypes, visualizations, dashboards, diagrams, options, or substantial reusable content, with a portable local runtime and no Anthropic API.","https:\u002F\u002Fgithub.com\u002Fzuzu59\u002Fz-skills\u002Fblob\u002Fmaster\u002Fskills\u002Fuse-artifacts\u002FSKILL.md",[],[]]