[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M70P0F19QJYCV9YY458YYGRF":3,"skill-same:01M70P0F19QJYCV9YY458YYGRF":78},{"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":34,"publishedAt":48,"linkOnly":49,"markdownContent":50,"markdownBody":51,"frontMatter":52,"refName":54,"commitSha":55,"skillDirectory":56,"forksCount":57,"repositoryHtmlUrl":58,"repositoryHomepageUrl":59,"ownerHtmlUrl":60,"ownerBlogUrl":61,"ownerDescription":62,"updatedAt":63,"install":64,"resources":77},"01M70P0F19QJYCV9YY458YYGRF","writing-agents",null,"How to write Claude Code agents that trigger reliably, use the right model, and produce consistent output. Use when creating, improving, or reviewing agent definitions.","shengdabai\u002FTony-Claude-Code-Skills","shengdabai","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F116417419?v=4","01M70F6D2WY5PQV5A92BKYJQEW","Tony Sheng","01M70F6D3GAAAYZ5H4CC1G8VXM",3,"MIT","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fxiaolai\u002Fnlpm\u002Fskills\u002Fnlpm\u002Fwriting-agents\u002FSKILL.md",[18,23,26,31],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000018","写作与文案","SCENARIO","使用场景",{"categoryNo":24,"categoryName":25,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发",{"categoryNo":27,"categoryName":28,"taxonomyType":29,"taxonomyTypeDesc":30},"01JBSEEDCAT000000000000030","开发工程师","PROFESSION","适用职业",{"categoryNo":32,"categoryName":33,"taxonomyType":29,"taxonomyTypeDesc":30},"01JBSEEDCAT000000000000036","内容创作者",[35,36,37,38,39,40,41,42,43,44,45,46,47],"writing","writing agents","model","how to","when","code","how","write","code agents","trigger","produce","use","output","2026-11-10 10:12:30",false,"---\nname: writing-agents\ndescription: \"How to write Claude Code agents that trigger reliably, use the right model, and produce consistent output. Use when creating, improving, or reviewing agent definitions.\"\nversion: 0.1.0\n---\n\n# Writing Agents\n\n> Scope: covers agent `.md` file authoring. For multi-agent orchestration, see [[orchestration]]. For plugin architecture, see [[writing-plugins]].\n\n## 1. Example Blocks Make or Break Triggering\n\nWithout `\u003Cexample>` blocks, Claude guesses when to dispatch your agent. With them, it pattern-matches against real scenarios.\n\n**Minimum**: 2 examples. **Ideal**: 3 -- one obvious trigger, one edge case, one non-obvious.\n\n### Example Block Anatomy\n\n```xml\n\u003Cexample>\nContext: [what the user is doing -- not just \"user needs help\"]\nuser: \"[realistic user message that should trigger this agent]\"\nassistant: \"[what Claude says when dispatching -- shows the decision logic]\"\n\u003C\u002Fexample>\n```\n\n### Bad Example (too vague -- 40% trigger accuracy)\n\n```xml\n\u003Cexample>\nContext: User needs code review\nuser: \"review my code\"\nassistant: \"I'll use the reviewer agent.\"\n\u003C\u002Fexample>\n```\n\n**Problems**: generic context, generic query, no decision logic shown.\n\n### Good Example (specific scenario -- 92% trigger accuracy)\n\n```xml\n\u003Cexample>\nContext: User just pushed changes to the authentication module and wants feedback before merging\nuser: \"Can you check if the auth changes look good before I create the PR?\"\nassistant: \"I'll dispatch the security-reviewer agent to check the auth changes for vulnerabilities, token handling, and session management best practices.\"\n\u003C\u002Fexample>\n```\n\n**Why it works**: specific context (auth module, pre-PR), realistic query (how users actually talk), decision logic visible (what the agent will check).\n\n### The Three-Example Pattern\n\n| Example | Purpose | What it demonstrates |\n|---------|---------|---------------------|\n| 1. Obvious trigger | Baseline dispatch | User explicitly asks for what the agent does |\n| 2. Edge case | Boundary behavior | User asks something adjacent -- agent should still trigger |\n| 3. Non-obvious | Discovery | User doesn't know the agent exists but their need matches |\n\nExample for a \"performance-profiler\" agent:\n\n```xml\n\u003C!-- Example 1: Obvious -->\n\u003Cexample>\nContext: User wants to profile their API\nuser: \"Profile the \u002Fapi\u002Fusers endpoint, it's slow\"\nassistant: \"I'll dispatch the performance-profiler to trace the \u002Fapi\u002Fusers endpoint...\"\n\u003C\u002Fexample>\n\n\u003C!-- Example 2: Edge case -->\n\u003Cexample>\nContext: User notices high memory usage but doesn't mention profiling\nuser: \"The app uses 2GB of RAM after running for an hour, is there a leak?\"\nassistant: \"I'll dispatch the performance-profiler to analyze memory allocation patterns...\"\n\u003C\u002Fexample>\n\n\u003C!-- Example 3: Non-obvious -->\n\u003Cexample>\nContext: User is comparing two implementation approaches\nuser: \"Should I use a JOIN here or two separate queries?\"\nassistant: \"I'll dispatch the performance-profiler to benchmark both approaches...\"\n\u003C\u002Fexample>\n```\n\n## 2. Model Selection\n\n| Task type | Model | Signal words in body | Examples |\n|-----------|-------|---------------------|----------|\n| Mechanical \u002F parsing \u002F formatting \u002F counting | haiku | count, list, extract, format, parse, scan | scanner, parser, formatter, counter, lister |\n| Analysis \u002F reasoning \u002F moderate judgment | sonnet | analyze, review, evaluate, summarize, compare | linter, reviewer, extractor, summarizer |\n| Complex judgment \u002F orchestration \u002F multi-agent | opus | coordinate, decide, assess, synthesize, architect | QC coordinator, architect, strategy planner |\n\n### Quick Heuristic\n\nCount the instruction lines in your agent body. Then check for judgment words.\n\n```\n\u003C 20 instruction lines AND no judgment words → haiku\n20-50 instruction lines OR judgment words → sonnet\n> 50 instruction lines OR coordination logic → opus\n```\n\nJudgment words: evaluate, decide, assess, determine, weigh, prioritize, recommend, judge, infer, synthesize.\n\n### Cost Impact\n\n| Model | Relative cost | When to upgrade |\n|-------|--------------|-----------------|\n| haiku | 1x | Agent produces wrong output on edge cases |\n| sonnet | 10x | Agent produces wrong output on easy cases |\n| opus | 30x | Agent coordinates other agents or makes architectural decisions |\n\n**Rule**: start with haiku, upgrade only when output quality requires it.\n\n## 3. Tool Least-Privilege\n\nOnly list tools the agent body actually references. Every extra tool is a potential misuse vector.\n\n### Common Mistakes\n\n| Agent type | Common over-grant | Correct tools |\n|-----------|-------------------|---------------|\n| Audit\u002Freview agent | Write, Edit, Bash | Read, Glob, Grep |\n| Code generator | Read, Grep (unused) | Write, Edit, Bash |\n| Orchestrator | Read, Write (does no IO) | Task |\n| Scanner | Bash (uses grep) | Grep, Glob, Read |\n\n### Tool Reference\n\n| Tool | When to include |\n|------|----------------|\n| Read | Agent reads file contents |\n| Write | Agent creates new files |\n| Edit | Agent modifies existing files |\n| Glob | Agent searches for files by pattern |\n| Grep | Agent searches file contents |\n| Bash | Agent runs shell commands (linters, tests, builds) |\n| Task | Agent dispatches sub-agents |\n| Fetch | Agent makes HTTP requests |\n\n## 4. Output Format\n\nEvery agent MUST define its output format in the body. Without it, output varies between invocations -- making results unparseable by parent agents.\n\n### Pattern: Structured Report\n\n```markdown\n## Output Format\n\n### {Section Title}\n| Column1 | Column2 | Column3 |\n|---------|---------|---------|\n| ...     | ...     | ...     |\n\n### Summary\n- Total items: {N}\n- Issues found: {N}\n- Pass\u002FFail: {verdict}\n```\n\n### Pattern: Severity-Tagged Findings\n\n```markdown\n## Output Format\n\nFor each finding, output:\n\n**[SEVERITY] Finding title**\n- File: `path\u002Fto\u002Ffile`\n- Line: {N}\n- Issue: {description}\n- Fix: {concrete suggestion}\n\nSeverity levels: CRITICAL > HIGH > MEDIUM > LOW > INFO\n```\n\n### Pattern: Pass\u002FFail Gate\n\n```markdown\n## Output Format\n\nFinal line must be exactly one of:\n- `PASS: All checks passed`\n- `WARN: {N} warnings found (see above)`\n- `FAIL: {N} errors found (see above)`\n```\n\n## 5. System Prompt Structure\n\nOrder matters. Claude reads top-to-bottom and front-loads early instructions.\n\n### The Five Sections\n\n```markdown\n## Mission\n[1-2 sentences: what this agent does and WHY it exists]\n\n## Instructions\n1. [First step]\n2. [Second step]\n3. [Third step]\n...\n\n## Boundaries\n- Do NOT [thing that would be harmful]\n- Do NOT [thing that's out of scope]\n- If [ambiguous situation], then [explicit resolution]\n\n## Output Format\n[Exact template -- see section 4 above]\n\n## Error Handling\n- If no files found: report \"No matching files\" and exit\n- If tool fails: report the error and continue with remaining work\n- If scope is unclear: analyze the narrower interpretation\n```\n\n### Section Sizing\n\n| Section | Target lines | Over-budget signal |\n|---------|-------------|-------------------|\n| Mission | 2-3 | More than one paragraph |\n| Instructions | 5-15 | More than 20 numbered steps |\n| Boundaries | 3-7 | More than 10 \"Do NOT\" items |\n| Output Format | 5-15 | Defining more than 3 output sections |\n| Error Handling | 3-5 | More than 5 error cases |\n\nTotal agent body: aim for 25-45 lines. Over 60 lines means the agent is doing too much -- split it.\n\n## 6. Worked Example\n\n### Before (score 52\u002F100)\n\n```yaml\n---\nname: code-checker\ndescription: Check code\nmodel: opus\ntools: [Read, Write, Edit, Bash, Grep, Glob, Task]\n---\n```\n\n```markdown\nYou are a code checker. Check the user's code for issues.\nLook at the files and find problems. Report what you find.\n```\n\n**Problems:**\n- Description: 0 trigger phrases, no \"Use when...\" (-30)\n- Model: opus for a simple review task (-10)\n- Tools: 7 tools granted, body uses maybe 3 (-10)\n- No examples: unreliable triggering (-15)\n- No output format: inconsistent results (-15)\n- No boundaries: scope creep (-10)\n- No error handling: silent failures (-10)\n\n### After (score 95\u002F100)\n\n```yaml\n---\nname: code-checker\ndescription: \"Static analysis agent — checks code for bugs, type errors, and anti-patterns. Use when reviewing code quality, running pre-commit checks, or validating changes before PR.\"\nmodel: sonnet\ntools: [Read, Glob, Grep]\n---\n```\n\n```markdown\n## Mission\nAnalyze source code files for bugs, type errors, and anti-patterns.\nProduce a structured report with severity-tagged findings.\n\n## Instructions\n1. Use Glob to discover files matching the target pattern\n2. Use Read to examine each file\n3. Use Grep to cross-reference imports and usage patterns\n4. For each issue found, classify severity and provide a concrete fix\n5. Produce the output report\n\n## Boundaries\n- Do NOT modify any files (read-only analysis)\n- Do NOT run shell commands\n- Do NOT report style issues (defer to linter)\n- If no target pattern specified, analyze all files in src\u002F\n\n## Output Format\n\nFor each finding:\n\n**[SEVERITY] Issue title**\n- File: `path\u002Fto\u002Ffile`\n- Line: {N}\n- Issue: {description}\n- Fix: {concrete fix}\n\nFinal line:\n- `PASS: No issues found`\n- `WARN: {N} warnings found`\n- `FAIL: {N} errors found`\n\n## Error Handling\n- If no files match the pattern: report \"No matching files for pattern: {X}\"\n- If a file cannot be read: skip it and note in the report\n```\n\n```xml\n\u003Cexample>\nContext: User just finished implementing a new feature and wants a quality check\nuser: \"Check the auth module for any bugs before I push\"\nassistant: \"I'll dispatch the code-checker agent to analyze src\u002Fauth\u002F for bugs, type errors, and anti-patterns.\"\n\u003C\u002Fexample>\n\n\u003Cexample>\nContext: User is debugging a production issue and suspects a code defect\nuser: \"Something's wrong with the payment flow, can you scan it?\"\nassistant: \"I'll dispatch the code-checker to analyze the payment module for potential bugs and logic errors.\"\n\u003C\u002Fexample>\n```\n\n**Changes made:**\n1. Description: 0 -> 6 trigger phrases (+30)\n2. Model: opus -> sonnet (appropriate for analysis, -20x cost) (+10)\n3. Tools: 7 -> 3 (read-only analysis needs read-only tools) (+10)\n4. Added 2 examples (+15)\n5. Defined output format (+15)\n6. Added boundaries (+10)\n7. Added error handling (+5)\n\n## 7. Common Mistakes\n\n| Mistake | Impact | Fix |\n|---------|--------|-----|\n| No examples | 40% trigger accuracy | Add 2-3 specific scenario examples |\n| Opus for mechanical work | 30x cost for same result | Use haiku for parsing, sonnet for analysis |\n| All tools granted | Agent writes when it should only read | List only tools the body references |\n| No output format | Different format each run | Define exact output template |\n| Body over 60 lines | Agent is doing too much | Split into focused sub-agents |\n| \"Be thorough\" in body | Meaningless filler | Replace with specific instructions |\n| No error handling | Silent failures | Add 3-5 error cases with resolution |\n","# Writing Agents\n\n> Scope: covers agent `.md` file authoring. For multi-agent orchestration, see [[orchestration]]. For plugin architecture, see [[writing-plugins]].\n\n## 1. Example Blocks Make or Break Triggering\n\nWithout `\u003Cexample>` blocks, Claude guesses when to dispatch your agent. With them, it pattern-matches against real scenarios.\n\n**Minimum**: 2 examples. **Ideal**: 3 -- one obvious trigger, one edge case, one non-obvious.\n\n### Example Block Anatomy\n\n```xml\n\u003Cexample>\nContext: [what the user is doing -- not just \"user needs help\"]\nuser: \"[realistic user message that should trigger this agent]\"\nassistant: \"[what Claude says when dispatching -- shows the decision logic]\"\n\u003C\u002Fexample>\n```\n\n### Bad Example (too vague -- 40% trigger accuracy)\n\n```xml\n\u003Cexample>\nContext: User needs code review\nuser: \"review my code\"\nassistant: \"I'll use the reviewer agent.\"\n\u003C\u002Fexample>\n```\n\n**Problems**: generic context, generic query, no decision logic shown.\n\n### Good Example (specific scenario -- 92% trigger accuracy)\n\n```xml\n\u003Cexample>\nContext: User just pushed changes to the authentication module and wants feedback before merging\nuser: \"Can you check if the auth changes look good before I create the PR?\"\nassistant: \"I'll dispatch the security-reviewer agent to check the auth changes for vulnerabilities, token handling, and session management best practices.\"\n\u003C\u002Fexample>\n```\n\n**Why it works**: specific context (auth module, pre-PR), realistic query (how users actually talk), decision logic visible (what the agent will check).\n\n### The Three-Example Pattern\n\n| Example | Purpose | What it demonstrates |\n|---------|---------|---------------------|\n| 1. Obvious trigger | Baseline dispatch | User explicitly asks for what the agent does |\n| 2. Edge case | Boundary behavior | User asks something adjacent -- agent should still trigger |\n| 3. Non-obvious | Discovery | User doesn't know the agent exists but their need matches |\n\nExample for a \"performance-profiler\" agent:\n\n```xml\n\u003C!-- Example 1: Obvious -->\n\u003Cexample>\nContext: User wants to profile their API\nuser: \"Profile the \u002Fapi\u002Fusers endpoint, it's slow\"\nassistant: \"I'll dispatch the performance-profiler to trace the \u002Fapi\u002Fusers endpoint...\"\n\u003C\u002Fexample>\n\n\u003C!-- Example 2: Edge case -->\n\u003Cexample>\nContext: User notices high memory usage but doesn't mention profiling\nuser: \"The app uses 2GB of RAM after running for an hour, is there a leak?\"\nassistant: \"I'll dispatch the performance-profiler to analyze memory allocation patterns...\"\n\u003C\u002Fexample>\n\n\u003C!-- Example 3: Non-obvious -->\n\u003Cexample>\nContext: User is comparing two implementation approaches\nuser: \"Should I use a JOIN here or two separate queries?\"\nassistant: \"I'll dispatch the performance-profiler to benchmark both approaches...\"\n\u003C\u002Fexample>\n```\n\n## 2. Model Selection\n\n| Task type | Model | Signal words in body | Examples |\n|-----------|-------|---------------------|----------|\n| Mechanical \u002F parsing \u002F formatting \u002F counting | haiku | count, list, extract, format, parse, scan | scanner, parser, formatter, counter, lister |\n| Analysis \u002F reasoning \u002F moderate judgment | sonnet | analyze, review, evaluate, summarize, compare | linter, reviewer, extractor, summarizer |\n| Complex judgment \u002F orchestration \u002F multi-agent | opus | coordinate, decide, assess, synthesize, architect | QC coordinator, architect, strategy planner |\n\n### Quick Heuristic\n\nCount the instruction lines in your agent body. Then check for judgment words.\n\n```\n\u003C 20 instruction lines AND no judgment words → haiku\n20-50 instruction lines OR judgment words → sonnet\n> 50 instruction lines OR coordination logic → opus\n```\n\nJudgment words: evaluate, decide, assess, determine, weigh, prioritize, recommend, judge, infer, synthesize.\n\n### Cost Impact\n\n| Model | Relative cost | When to upgrade |\n|-------|--------------|-----------------|\n| haiku | 1x | Agent produces wrong output on edge cases |\n| sonnet | 10x | Agent produces wrong output on easy cases |\n| opus | 30x | Agent coordinates other agents or makes architectural decisions |\n\n**Rule**: start with haiku, upgrade only when output quality requires it.\n\n## 3. Tool Least-Privilege\n\nOnly list tools the agent body actually references. Every extra tool is a potential misuse vector.\n\n### Common Mistakes\n\n| Agent type | Common over-grant | Correct tools |\n|-----------|-------------------|---------------|\n| Audit\u002Freview agent | Write, Edit, Bash | Read, Glob, Grep |\n| Code generator | Read, Grep (unused) | Write, Edit, Bash |\n| Orchestrator | Read, Write (does no IO) | Task |\n| Scanner | Bash (uses grep) | Grep, Glob, Read |\n\n### Tool Reference\n\n| Tool | When to include |\n|------|----------------|\n| Read | Agent reads file contents |\n| Write | Agent creates new files |\n| Edit | Agent modifies existing files |\n| Glob | Agent searches for files by pattern |\n| Grep | Agent searches file contents |\n| Bash | Agent runs shell commands (linters, tests, builds) |\n| Task | Agent dispatches sub-agents |\n| Fetch | Agent makes HTTP requests |\n\n## 4. Output Format\n\nEvery agent MUST define its output format in the body. Without it, output varies between invocations -- making results unparseable by parent agents.\n\n### Pattern: Structured Report\n\n```markdown\n## Output Format\n\n### {Section Title}\n| Column1 | Column2 | Column3 |\n|---------|---------|---------|\n| ...     | ...     | ...     |\n\n### Summary\n- Total items: {N}\n- Issues found: {N}\n- Pass\u002FFail: {verdict}\n```\n\n### Pattern: Severity-Tagged Findings\n\n```markdown\n## Output Format\n\nFor each finding, output:\n\n**[SEVERITY] Finding title**\n- File: `path\u002Fto\u002Ffile`\n- Line: {N}\n- Issue: {description}\n- Fix: {concrete suggestion}\n\nSeverity levels: CRITICAL > HIGH > MEDIUM > LOW > INFO\n```\n\n### Pattern: Pass\u002FFail Gate\n\n```markdown\n## Output Format\n\nFinal line must be exactly one of:\n- `PASS: All checks passed`\n- `WARN: {N} warnings found (see above)`\n- `FAIL: {N} errors found (see above)`\n```\n\n## 5. System Prompt Structure\n\nOrder matters. Claude reads top-to-bottom and front-loads early instructions.\n\n### The Five Sections\n\n```markdown\n## Mission\n[1-2 sentences: what this agent does and WHY it exists]\n\n## Instructions\n1. [First step]\n2. [Second step]\n3. [Third step]\n...\n\n## Boundaries\n- Do NOT [thing that would be harmful]\n- Do NOT [thing that's out of scope]\n- If [ambiguous situation], then [explicit resolution]\n\n## Output Format\n[Exact template -- see section 4 above]\n\n## Error Handling\n- If no files found: report \"No matching files\" and exit\n- If tool fails: report the error and continue with remaining work\n- If scope is unclear: analyze the narrower interpretation\n```\n\n### Section Sizing\n\n| Section | Target lines | Over-budget signal |\n|---------|-------------|-------------------|\n| Mission | 2-3 | More than one paragraph |\n| Instructions | 5-15 | More than 20 numbered steps |\n| Boundaries | 3-7 | More than 10 \"Do NOT\" items |\n| Output Format | 5-15 | Defining more than 3 output sections |\n| Error Handling | 3-5 | More than 5 error cases |\n\nTotal agent body: aim for 25-45 lines. Over 60 lines means the agent is doing too much -- split it.\n\n## 6. Worked Example\n\n### Before (score 52\u002F100)\n\n```yaml\n---\nname: code-checker\ndescription: Check code\nmodel: opus\ntools: [Read, Write, Edit, Bash, Grep, Glob, Task]\n---\n```\n\n```markdown\nYou are a code checker. Check the user's code for issues.\nLook at the files and find problems. Report what you find.\n```\n\n**Problems:**\n- Description: 0 trigger phrases, no \"Use when...\" (-30)\n- Model: opus for a simple review task (-10)\n- Tools: 7 tools granted, body uses maybe 3 (-10)\n- No examples: unreliable triggering (-15)\n- No output format: inconsistent results (-15)\n- No boundaries: scope creep (-10)\n- No error handling: silent failures (-10)\n\n### After (score 95\u002F100)\n\n```yaml\n---\nname: code-checker\ndescription: \"Static analysis agent — checks code for bugs, type errors, and anti-patterns. Use when reviewing code quality, running pre-commit checks, or validating changes before PR.\"\nmodel: sonnet\ntools: [Read, Glob, Grep]\n---\n```\n\n```markdown\n## Mission\nAnalyze source code files for bugs, type errors, and anti-patterns.\nProduce a structured report with severity-tagged findings.\n\n## Instructions\n1. Use Glob to discover files matching the target pattern\n2. Use Read to examine each file\n3. Use Grep to cross-reference imports and usage patterns\n4. For each issue found, classify severity and provide a concrete fix\n5. Produce the output report\n\n## Boundaries\n- Do NOT modify any files (read-only analysis)\n- Do NOT run shell commands\n- Do NOT report style issues (defer to linter)\n- If no target pattern specified, analyze all files in src\u002F\n\n## Output Format\n\nFor each finding:\n\n**[SEVERITY] Issue title**\n- File: `path\u002Fto\u002Ffile`\n- Line: {N}\n- Issue: {description}\n- Fix: {concrete fix}\n\nFinal line:\n- `PASS: No issues found`\n- `WARN: {N} warnings found`\n- `FAIL: {N} errors found`\n\n## Error Handling\n- If no files match the pattern: report \"No matching files for pattern: {X}\"\n- If a file cannot be read: skip it and note in the report\n```\n\n```xml\n\u003Cexample>\nContext: User just finished implementing a new feature and wants a quality check\nuser: \"Check the auth module for any bugs before I push\"\nassistant: \"I'll dispatch the code-checker agent to analyze src\u002Fauth\u002F for bugs, type errors, and anti-patterns.\"\n\u003C\u002Fexample>\n\n\u003Cexample>\nContext: User is debugging a production issue and suspects a code defect\nuser: \"Something's wrong with the payment flow, can you scan it?\"\nassistant: \"I'll dispatch the code-checker to analyze the payment module for potential bugs and logic errors.\"\n\u003C\u002Fexample>\n```\n\n**Changes made:**\n1. Description: 0 -> 6 trigger phrases (+30)\n2. Model: opus -> sonnet (appropriate for analysis, -20x cost) (+10)\n3. Tools: 7 -> 3 (read-only analysis needs read-only tools) (+10)\n4. Added 2 examples (+15)\n5. Defined output format (+15)\n6. Added boundaries (+10)\n7. Added error handling (+5)\n\n## 7. Common Mistakes\n\n| Mistake | Impact | Fix |\n|---------|--------|-----|\n| No examples | 40% trigger accuracy | Add 2-3 specific scenario examples |\n| Opus for mechanical work | 30x cost for same result | Use haiku for parsing, sonnet for analysis |\n| All tools granted | Agent writes when it should only read | List only tools the body references |\n| No output format | Different format each run | Define exact output template |\n| Body over 60 lines | Agent is doing too much | Split into focused sub-agents |\n| \"Be thorough\" in body | Meaningless filler | Replace with specific instructions |\n| No error handling | Silent failures | Add 3-5 error cases with resolution |\n",{"name":5,"version":53,"description":7},"0.1.0","main","7850740bd837adb32f0a109ab2af82482196c6f2","skills\u002Fxiaolai\u002Fnlpm\u002Fskills\u002Fnlpm\u002Fwriting-agents",1,"https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills","https:\u002F\u002Fzturnsgo.com","https:\u002F\u002Fgithub.com\u002Fshengdabai","https:\u002F\u002Fzturnsgo.com\u002F","AI × education creator — producer, not consumer. Taught 6,000+ learners; shipping learning tools & daily essays with Claude Code, practicing @xiaolai's way 🥋","2026-11-01 21:28:23",{"source":8,"skillName":5,"commands":65,"prompt":75,"zipAvailable":76},[66,69,72],{"tool":67,"command":68},"npx","npx skills add shengdabai\u002FTony-Claude-Code-Skills --skill writing-agents",{"tool":70,"command":71},"bunx","bunx skills add shengdabai\u002FTony-Claude-Code-Skills --skill writing-agents",{"tool":73,"command":74},"pnpm","pnpm dlx skills add shengdabai\u002FTony-Claude-Code-Skills --skill writing-agents","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fxiaolai\u002Fnlpm\u002Fskills\u002Fnlpm\u002Fwriting-agents\u002FSKILL.md",true,[],[79,88,96,104,112,120],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":83,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":87},"01M73H7XKBHE1ETQDZT94MF9HH","seedance-prompt-en","Write effective prompts for Jimeng Seedance 2.0 multimodal AI video generation. Use when users want to create video prompts using text, images, videos, and audio inputs with the @ reference system. Covers camera movements, effects replication, video extension, editing, music beat-matching, e-commerce ads, short dramas, and educational content.","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fseedance-prompt-en\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fseedance-prompt-en\u002FSKILL.md",[],[],"2026-11-11 15:11:14",{"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":92,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":87},"01M73GRE9998Q1DZWC71N41JWG","onepage-pdf","Convert an HTML page into a single continuous-page PDF (one tall page, no pagination breaks), preserving desktop layout, backgrounds and selectable text. Use when the user asks to turn an HTML file\u002Freport\u002Fproposal into a \"single-page PDF\", \"long PDF\", \"不分页 PDF\", \"单页 PDF\", \"长图式 PDF\", or complains that an HTML-to-PDF export breaks into pages or cuts content. Supports optional string redaction with leak verification.","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fonepage-pdf\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fonepage-pdf\u002FSKILL.md",[],[],{"skillNo":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":100,"sourceHtmlUrl":101,"categories":102,"keywords":103,"publishedAt":87},"01M73GQG1KZDADVPV3MHAT1J5A","native-feel-cross-platform-desktop","Use when the user is designing, prototyping, or rewriting a desktop app that must run on multiple OSes (macOS + Windows, optionally Linux) AND feel indistinguishable from a native app to its users — fast launch, native windowing, native input handling, native materials. Trigger words include \"cross-platform desktop\", \"Electron alternative\", \"Tauri vs native\", \"WebView wrapper\", \"near-native performance\", \"Raycast architecture\", \"WebKit\u002FWebView2 quirks\", \"WKWebView\", \"system tray app\", \"global hotkey app\", \"launcher app\". Do NOT trigger this skill for pure web apps, pure mobile apps, or for greenfield projects that have no native-feel requirement.","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fnative-feel-cross-platform-desktop\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fnative-feel-cross-platform-desktop\u002FSKILL.md",[],[],{"skillNo":105,"name":106,"nameZh":6,"description":107,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":108,"sourceHtmlUrl":109,"categories":110,"keywords":111,"publishedAt":87},"01M73G87PDNSMKG2FHWCYER6N8","gstack","Router for the gstack skill suite. (gstack)","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fgstack\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fgstack\u002FSKILL.md",[],[],{"skillNo":113,"name":114,"nameZh":6,"description":115,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":116,"sourceHtmlUrl":117,"categories":118,"keywords":119,"publishedAt":87},"01M73G3EAQWKCM8DB8D0N0QSVG","fireworks-tech-graph","Use when the user wants to create any technical diagram - architecture, data flow, flowchart, sequence, agent\u002Fmemory, or concept map - and export as SVG+PNG. Trigger on: \"画图\" \"帮我画\" \"生成图\" \"做个图\" \"架构图\" \"流程图\" \"可视化一下\" \"出图\" \"generate diagram\" \"draw diagram\" \"visualize\" or any system\u002Fflow description the user wants illustrated.","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Ffireworks-tech-graph\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Ffireworks-tech-graph\u002FSKILL.md",[],[],{"skillNo":121,"name":122,"nameZh":6,"description":123,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":124,"categories":125,"keywords":126,"publishedAt":87},"01M73G086AXAH9DW1H5G5N3CPB","dbs-video-extract","接收抖音、小红书、微信视频号等短视频链接或分享文案，查询作品／账号数据并提取语音文字稿，按作者和标题归档为 Markdown。用户希望完整解析短视频、取得数据或文字稿时使用；运行前检查 TikHub 与轻抖使用凭证，缺少时用新手能理解的方式引导购买和安全配置。","https:\u002F\u002Fgithub.com\u002Fshengdabai\u002FTony-Claude-Code-Skills\u002Fblob\u002Fmain\u002Fskills\u002Fdbs-video-extract\u002FSKILL.md",[],[]]