[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6CD24MNB8JF63XFKXX2ZX8Z":3,"skill-same:01M6CD24MNB8JF63XFKXX2ZX8Z":86},{"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":23,"publishedAt":41,"linkOnly":42,"markdownContent":43,"markdownBody":44,"frontMatter":45,"refName":52,"commitSha":53,"skillDirectory":54,"forksCount":14,"repositoryHtmlUrl":55,"repositoryHomepageUrl":6,"ownerHtmlUrl":56,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":57,"install":58,"resources":71},"01M6CD24MNB8JF63XFKXX2ZX8Z","prompt-engineer",null,"Optimize system prompts for Claude Code agents using proven prompt engineering patterns. Use when users request prompt improvement, optimization, or refinement for agent workflows, tool instructions, or system behaviors.","loriensleafs\u002Fbrain","loriensleafs","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F315385?v=4","01M6CCYZETMCJPTHNS258YRCBC","Peter Kloss","01M6CCYZK25T8516A39S9PCJT7",0,"MIT","https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fprompt-engineer\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40],"engineer","prompt","prompt engineer","prompt engineering patterns","prompts","use","instructions","tool","system","architect","prompt engineering","Optimization","engineering","code agents","code","engineering patterns","workflows","2026-11-10 10:12:30",false,"---\nname: prompt-engineer\ndescription: Optimize system prompts for Claude Code agents using proven prompt engineering patterns. Use when users request prompt improvement, optimization, or refinement for agent workflows, tool instructions, or system behaviors.\nlicense: MIT\nagents:\n  - architect\n  - analyst\nmodel: claude-opus-4-6[1m]\nmetadata:\n  version: 1.0.0\n  timelessness: 9\u002F10\n---\n\n# Prompt Optimizer\n\nOptimizes system prompts by applying research-backed prompt engineering patterns. Human-in-the-loop phases: understand, plan, propose changes, receive approval, then integrate.\n\n## Purpose and Success Criteria\n\nA well-optimized prompt achieves:\n\n1. **Behavioral clarity**: Agent knows exactly what to do in common cases and edge cases\n2. **Appropriate scope**: Complex tasks get decomposition; simple tasks don't trigger overthinking\n3. **Grounded changes**: Every modification traces to a specific pattern with documented impact\n\nOptimization is complete when:\n\n- Every change has explicit pattern attribution from the reference document\n- No section contradicts another section\n- The prompt matches its operating context (tool-use vs. conversational, token constraints)\n- Human has approved both section-level changes and full integration\n\n## Triggers\n\n| Trigger Phrase                          | Operation                                     |\n| --------------------------------------- | --------------------------------------------- |\n| `optimize this prompt`                  | Full Phase 0-4 optimization workflow          |\n| `improve this system prompt`            | Analyze and propose changes with visual cards |\n| `review my agent prompt`                | Pattern-based review against reference        |\n| `refine this prompt for better results` | Targeted improvement with BEFORE\u002FAFTER        |\n| `make this prompt more effective`       | Technique selection and application           |\n\n---\n\n## When to Use This Skill\n\nUse when the user provides a prompt and wants it improved, refined, or reviewed for best practices.\n\nDo NOT use for:\n\n- Writing prompts from scratch (different skill)\n- Prompts that are already working well and user just wants validation\n- Non-prompt content (documentation, code, etc.)\n\n## Required Resources\n\nBefore ANY analysis, read the appropriate pattern reference(s):\n\n### Single-Turn Reference (Always Read)\n\n```text\nRead references\u002Fprompt-engineering-single-turn.md\n```\n\nContains: Technique Selection Guide table, Quick Reference principles, domain-organized techniques with citations, Anti-Patterns section.\n\n### Multi-Turn Reference (Conditional)\n\n```text\nRead references\u002Fprompt-engineering-multi-turn.md\n```\n\n**Read ONLY when the prompt involves:**\n\n- Multi-turn flows (iterative refinement, conversation chains)\n- Multi-agent \u002F sub-agent orchestration\n\n**Skip for:**\n\n- Static system prompts executed in a single LLM call\n- Tool instructions or one-shot prompts\n\n### Workflow Reference\n\n```text\nRead references\u002Fworkflow.md\n```\n\nContains: Detailed Phase 0-4 workflows, visual card template, completion checkpoint.\n\n## Process\n\n```text\n┌─────────────────────────────────────────────────────────────────┐\n│ 1. READ THE REFERENCE(S)                                        │\n│    - Always: references\u002Fprompt-engineering-single-turn.md       │\n│    - If multi-turn\u002Fmulti-agent: also read multi-turn reference  │\n├─────────────────────────────────────────────────────────────────┤\n│ 2. UNDERSTAND THE PROMPT (Phase 1)                              │\n│    - Operating context (single-shot? tool-use? constraints?)    │\n│    - Current state (working? unclear? missing?)                 │\n│    - Document specific problems with quoted prompt text         │\n├─────────────────────────────────────────────────────────────────┤\n│ 3. PLAN WITH VISUAL CARDS (Phase 2)                             │\n│    - Present each change as a visual card with:                 │\n│      SCOPE → PROBLEM → TECHNIQUE → BEFORE\u002FAFTER                 │\n│    - Quote trigger conditions from reference                    │\n│    - ⚠️  WAIT FOR USER APPROVAL before proceeding               │\n├─────────────────────────────────────────────────────────────────┤\n│ 4. EXECUTE APPROVED CHANGES (Phase 3)                           │\n│    - Apply the BEFORE → AFTER transformations                   │\n├─────────────────────────────────────────────────────────────────┤\n│ 5. INTEGRATE AND VERIFY QUALITY (Phase 4)                       │\n│    - Check cross-section coherence                              │\n│    - Final anti-pattern check                                   │\n│    - Present complete optimized prompt                          │\n└─────────────────────────────────────────────────────────────────┘\n```\n\n## Triage (Phase 0)\n\n**Simple prompts** (use lightweight process):\n\n- Under 20 lines\n- Single clear purpose\n- No conditional logic\n\n**Complex prompts** (use full process):\n\n- Multiple sections serving different functions\n- Conditional behaviors or rule hierarchies\n- Tool orchestration or multi-step workflows\n\n## Core Quality Principles\n\n1. **Quote before deciding**: Every technique selection must quote the reference's trigger condition.\n2. **Open verification questions**: Ask \"What behavior will this produce?\" not \"Is this correct?\"\n3. **Approval happens once, upfront**: The visual card format in Phase 2 shows full impact.\n4. **Preserve what works**: Optimization means improving problems, not rewriting everything.\n\n## Completion Checkpoint\n\nBefore presenting the final prompt, verify:\n\n- [ ] Phase 2 plan used visual card format with BEFORE\u002FAFTER\n- [ ] Phase 2 plan quoted trigger conditions from reference\n- [ ] Phase 2 plan was approved by user before Phase 3\n- [ ] No technique applied without matching trigger condition\n- [ ] Stacking compatibility checked; no conflicts\n- [ ] Anti-patterns section consulted; none introduced\n- [ ] Emphasis markers used sparingly (≤3 highest-level)\n\n## Anti-Patterns\n\n| Avoid                                               | Why                                          | Instead                                            |\n| --------------------------------------------------- | -------------------------------------------- | -------------------------------------------------- |\n| Applying techniques without reading reference first | Missing trigger conditions and constraints   | Always read reference documents before analysis    |\n| Rewriting entire prompt                             | Destroys what already works                  | Preserve working sections, improve problems only   |\n| Skipping user approval before changes               | May misidentify improvement priorities       | Present visual cards in Phase 2, wait for approval |\n| Stacking conflicting techniques                     | Produces contradictory instructions          | Check stacking compatibility per reference         |\n| Using more than 3 emphasis markers                  | Dilutes signal when everything is emphasized | Reserve emphasis for highest-priority instructions |\n\n---\n\n## Verification\n\nAfter optimization:\n\n- [ ] Every change has pattern attribution from reference document\n- [ ] No section contradicts another section\n- [ ] User approved Phase 2 plan before Phase 3 execution\n- [ ] Anti-patterns section consulted, none introduced\n- [ ] Emphasis markers used sparingly (3 or fewer)\n\n---\n\n## References\n\n- [prompt-engineering-single-turn.md](references\u002Fprompt-engineering-single-turn.md) - Single-turn patterns\n- [prompt-engineering-multi-turn.md](references\u002Fprompt-engineering-multi-turn.md) - Multi-turn patterns\n- [workflow.md](references\u002Fworkflow.md) - Detailed phase workflows and card template\n","# Prompt Optimizer\n\nOptimizes system prompts by applying research-backed prompt engineering patterns. Human-in-the-loop phases: understand, plan, propose changes, receive approval, then integrate.\n\n## Purpose and Success Criteria\n\nA well-optimized prompt achieves:\n\n1. **Behavioral clarity**: Agent knows exactly what to do in common cases and edge cases\n2. **Appropriate scope**: Complex tasks get decomposition; simple tasks don't trigger overthinking\n3. **Grounded changes**: Every modification traces to a specific pattern with documented impact\n\nOptimization is complete when:\n\n- Every change has explicit pattern attribution from the reference document\n- No section contradicts another section\n- The prompt matches its operating context (tool-use vs. conversational, token constraints)\n- Human has approved both section-level changes and full integration\n\n## Triggers\n\n| Trigger Phrase                          | Operation                                     |\n| --------------------------------------- | --------------------------------------------- |\n| `optimize this prompt`                  | Full Phase 0-4 optimization workflow          |\n| `improve this system prompt`            | Analyze and propose changes with visual cards |\n| `review my agent prompt`                | Pattern-based review against reference        |\n| `refine this prompt for better results` | Targeted improvement with BEFORE\u002FAFTER        |\n| `make this prompt more effective`       | Technique selection and application           |\n\n---\n\n## When to Use This Skill\n\nUse when the user provides a prompt and wants it improved, refined, or reviewed for best practices.\n\nDo NOT use for:\n\n- Writing prompts from scratch (different skill)\n- Prompts that are already working well and user just wants validation\n- Non-prompt content (documentation, code, etc.)\n\n## Required Resources\n\nBefore ANY analysis, read the appropriate pattern reference(s):\n\n### Single-Turn Reference (Always Read)\n\n```text\nRead references\u002Fprompt-engineering-single-turn.md\n```\n\nContains: Technique Selection Guide table, Quick Reference principles, domain-organized techniques with citations, Anti-Patterns section.\n\n### Multi-Turn Reference (Conditional)\n\n```text\nRead references\u002Fprompt-engineering-multi-turn.md\n```\n\n**Read ONLY when the prompt involves:**\n\n- Multi-turn flows (iterative refinement, conversation chains)\n- Multi-agent \u002F sub-agent orchestration\n\n**Skip for:**\n\n- Static system prompts executed in a single LLM call\n- Tool instructions or one-shot prompts\n\n### Workflow Reference\n\n```text\nRead references\u002Fworkflow.md\n```\n\nContains: Detailed Phase 0-4 workflows, visual card template, completion checkpoint.\n\n## Process\n\n```text\n┌─────────────────────────────────────────────────────────────────┐\n│ 1. READ THE REFERENCE(S)                                        │\n│    - Always: references\u002Fprompt-engineering-single-turn.md       │\n│    - If multi-turn\u002Fmulti-agent: also read multi-turn reference  │\n├─────────────────────────────────────────────────────────────────┤\n│ 2. UNDERSTAND THE PROMPT (Phase 1)                              │\n│    - Operating context (single-shot? tool-use? constraints?)    │\n│    - Current state (working? unclear? missing?)                 │\n│    - Document specific problems with quoted prompt text         │\n├─────────────────────────────────────────────────────────────────┤\n│ 3. PLAN WITH VISUAL CARDS (Phase 2)                             │\n│    - Present each change as a visual card with:                 │\n│      SCOPE → PROBLEM → TECHNIQUE → BEFORE\u002FAFTER                 │\n│    - Quote trigger conditions from reference                    │\n│    - ⚠️  WAIT FOR USER APPROVAL before proceeding               │\n├─────────────────────────────────────────────────────────────────┤\n│ 4. EXECUTE APPROVED CHANGES (Phase 3)                           │\n│    - Apply the BEFORE → AFTER transformations                   │\n├─────────────────────────────────────────────────────────────────┤\n│ 5. INTEGRATE AND VERIFY QUALITY (Phase 4)                       │\n│    - Check cross-section coherence                              │\n│    - Final anti-pattern check                                   │\n│    - Present complete optimized prompt                          │\n└─────────────────────────────────────────────────────────────────┘\n```\n\n## Triage (Phase 0)\n\n**Simple prompts** (use lightweight process):\n\n- Under 20 lines\n- Single clear purpose\n- No conditional logic\n\n**Complex prompts** (use full process):\n\n- Multiple sections serving different functions\n- Conditional behaviors or rule hierarchies\n- Tool orchestration or multi-step workflows\n\n## Core Quality Principles\n\n1. **Quote before deciding**: Every technique selection must quote the reference's trigger condition.\n2. **Open verification questions**: Ask \"What behavior will this produce?\" not \"Is this correct?\"\n3. **Approval happens once, upfront**: The visual card format in Phase 2 shows full impact.\n4. **Preserve what works**: Optimization means improving problems, not rewriting everything.\n\n## Completion Checkpoint\n\nBefore presenting the final prompt, verify:\n\n- [ ] Phase 2 plan used visual card format with BEFORE\u002FAFTER\n- [ ] Phase 2 plan quoted trigger conditions from reference\n- [ ] Phase 2 plan was approved by user before Phase 3\n- [ ] No technique applied without matching trigger condition\n- [ ] Stacking compatibility checked; no conflicts\n- [ ] Anti-patterns section consulted; none introduced\n- [ ] Emphasis markers used sparingly (≤3 highest-level)\n\n## Anti-Patterns\n\n| Avoid                                               | Why                                          | Instead                                            |\n| --------------------------------------------------- | -------------------------------------------- | -------------------------------------------------- |\n| Applying techniques without reading reference first | Missing trigger conditions and constraints   | Always read reference documents before analysis    |\n| Rewriting entire prompt                             | Destroys what already works                  | Preserve working sections, improve problems only   |\n| Skipping user approval before changes               | May misidentify improvement priorities       | Present visual cards in Phase 2, wait for approval |\n| Stacking conflicting techniques                     | Produces contradictory instructions          | Check stacking compatibility per reference         |\n| Using more than 3 emphasis markers                  | Dilutes signal when everything is emphasized | Reserve emphasis for highest-priority instructions |\n\n---\n\n## Verification\n\nAfter optimization:\n\n- [ ] Every change has pattern attribution from reference document\n- [ ] No section contradicts another section\n- [ ] User approved Phase 2 plan before Phase 3 execution\n- [ ] Anti-patterns section consulted, none introduced\n- [ ] Emphasis markers used sparingly (3 or fewer)\n\n---\n\n## References\n\n- [prompt-engineering-single-turn.md](references\u002Fprompt-engineering-single-turn.md) - Single-turn patterns\n- [prompt-engineering-multi-turn.md](references\u002Fprompt-engineering-multi-turn.md) - Multi-turn patterns\n- [workflow.md](references\u002Fworkflow.md) - Detailed phase workflows and card template\n",{"name":5,"model":46,"agents":47,"license":15,"metadata":49,"description":7},"claude-opus-4-6[1m]",[33,48],"analyst",{"version":50,"timelessness":51},"1.0.0","9\u002F10","main","2bd10cb7ca989f88dbf338b57957763718eda59b","templates\u002Fskills\u002Fprompt-engineer","https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain","https:\u002F\u002Fgithub.com\u002Floriensleafs","2026-03-11 23:51:59",{"source":8,"skillName":5,"commands":59,"prompt":69,"zipAvailable":70},[60,63,66],{"tool":61,"command":62},"npx","npx skills add loriensleafs\u002Fbrain --skill prompt-engineer",{"tool":64,"command":65},"bunx","bunx skills add loriensleafs\u002Fbrain --skill prompt-engineer",{"tool":67,"command":68},"pnpm","pnpm dlx skills add loriensleafs\u002Fbrain --skill prompt-engineer","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fprompt-engineer\u002FSKILL.md",true,[72,78,82],{"path":73,"resourceType":74,"resourceTypeDesc":75,"sizeBytes":76,"sourceHtmlUrl":77},"references\u002Fprompt-engineering-multi-turn.md","REFERENCE","参考文档",29755,"https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fprompt-engineer\u002Freferences\u002Fprompt-engineering-multi-turn.md",{"path":79,"resourceType":74,"resourceTypeDesc":75,"sizeBytes":80,"sourceHtmlUrl":81},"references\u002Fprompt-engineering-single-turn.md",82176,"https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fprompt-engineer\u002Freferences\u002Fprompt-engineering-single-turn.md",{"path":83,"resourceType":74,"resourceTypeDesc":75,"sizeBytes":84,"sourceHtmlUrl":85},"references\u002Fworkflow.md",18782,"https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fprompt-engineer\u002Freferences\u002Fworkflow.md",[87,100,117,140,156,180],{"skillNo":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":41},"01M6CD3KHDHPR633RVP95G4XWN","validation-authority","Treat upstream validators as authoritative. Align local config to them. Use when validation fails unexpectedly, before modifying validator behavior, or when tempted to change upstream tool code.","https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fvalidation-authority\u002FSKILL.md",[],[94,95,96,29,97,98,31,33,38,99],"validation authority","validation","config","implementer","behavior","change",{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":41},"01M6CD3H29Q8FWN9YETT8YVQTS","using-brain-memory","Guidance for using Brain semantic memory effectively. Applies Zettelkasten atomic note principles with observations and relations. Use when deciding whether to query or create notes, structuring note content, or understanding the semantic knowledge graph.","https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fusing-forgetful-memory\u002FSKILL.md",[],[107,108,109,110,111,112,113,29,114,115,116],"using brain memory","memory","content","create","notes","graph","guidance","context","knowledge","knowledge graph",{"skillNo":118,"name":119,"nameZh":6,"description":120,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":121,"categories":122,"keywords":123,"publishedAt":41},"01M6CD3E2KTTMNSWXRJAAZA7D1","threat-modeling","Structured security analysis using OWASP Four-Question Framework and STRIDE methodology. Generates threat matrices with risk ratings, mitigations, and prioritization. Use for attack surface analysis, security architecture review, or when asking what can go wrong.","https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fthreat-modeling\u002FSKILL.md",[],[124,125,126,29,127,33,128,129,130,131,132,133,134,135,136,137,138,139],"threat","threat modeling","prioritization","question","architecture review","stride","analysis","risk","owasp","methodology","framework","security analysis","architecture","security","go","review",{"skillNo":141,"name":142,"nameZh":6,"description":143,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":144,"categories":145,"keywords":147,"publishedAt":41},"01M6CD3BC6G0Q2MDNW6X51XHBQ","taste-lints","Custom lints with agent-readable remediation instructions. Enforces taste invariants (file size, naming conventions, structured logging, complexity) and surfaces errors that agents can act on directly. Use when writing or reviewing code to catch style violations early.","https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Ftaste-lints\u002FSKILL.md",[146],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[148,97,149,150,29,30,151,152,153,154,38,155],"taste lints","surfaces","complexity","remediation","file","logging","qa","writing",{"skillNo":157,"name":158,"nameZh":6,"description":159,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":160,"categories":161,"keywords":162,"publishedAt":41},"01M6CD37Z7VD22BH42TG0GM66V","style-enforcement","Validate code against style rules from .editorconfig, StyleCop.json, and Directory.Build.props. Detects line ending violations, naming convention issues, indentation problems, and charset mismatches across C#, Python, TypeScript, and JavaScript. Produces JSON reports for pre-commit hooks and CI pipelines.","https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fstyle-enforcement\u002FSKILL.md",[],[163,164,165,166,167,168,97,169,170,171,172,173,154,174,38,175,176,177,178,179],"style enforcement","enforcement","ci","pipelines","directory","hooks","Rules","json","pre-commit","editorconfig","reports","issues","validate","commit","javascript","python","typescript",{"skillNo":181,"name":182,"nameZh":6,"description":183,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":184,"categories":185,"keywords":186,"publishedAt":41},"01M6CD35JW73TPMCZBF9DTDEE8","steering-matcher","Match file paths against steering file glob patterns to determine applicable steering guidance. Use when orchestrator needs to inject context-aware guidance based on files being modified.","https:\u002F\u002Fgithub.com\u002Floriensleafs\u002Fbrain\u002Fblob\u002Fmain\u002Ftemplates\u002Fskills\u002Fsteering-matcher\u002FSKILL.md",[],[187,29,188,189,114,190,113,152],"steering matcher","files","haiku","orchestrator"]