[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6HXYM3Y5CSG289ZQ5RHJ322":3,"skill-same:01M6HXYM3Y5CSG289ZQ5RHJ322":63},{"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":30,"linkOnly":31,"markdownContent":32,"markdownBody":33,"frontMatter":34,"refName":35,"commitSha":36,"skillDirectory":37,"forksCount":38,"repositoryHtmlUrl":39,"repositoryHomepageUrl":6,"ownerHtmlUrl":40,"ownerBlogUrl":41,"ownerDescription":6,"updatedAt":42,"install":43,"resources":56},"01M6HXYM3Y5CSG289ZQ5RHJ322","search-fail-pivot",null,"Heuristic for detecting search tool failures and pivoting to domain knowledge for document generation","HKUDS\u002FAgentPilot","HKUDS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F118165258?v=4","01M6HW8VRNZP64G0P4MVJW2KKQ","✨Data Intelligence Lab@HKU✨","01M6HW8VWZVHBRHYH3T330PFWV",8029,"MIT","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Flegacy\u002Fv1\u002Fgdpval_bench\u002Fskills\u002Fsearch-fail-pivot\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27,28,29],"search","search fail pivot","pivoting","knowledge","tool","document","domain","generation","document generation","detecting","for","2026-11-10 10:12:30",false,"---\nname: search-fail-pivot\ndescription: Heuristic for detecting search tool failures and pivoting to domain knowledge for document generation\n---\n\n# Search Tool Failure Detection and Pivot Strategy\n\n## Purpose\n\nWhen using `search_web` or similar external data retrieval tools, repeated failures waste iterations. This skill provides a diagnostic heuristic to recognize when to abandon search attempts and proceed with existing domain knowledge.\n\n## Core Heuristic\n\n**If `search_web` fails 2+ consecutive times with 'unknown error' or similar non-recoverable errors, pivot to domain knowledge + `run_shell` for document generation.**\n\nDo not attempt more than 2 retries on the same or similar queries before pivoting.\n\n## Step-by-Step Instructions\n\n### Step 1: Track Search Failures\n\nMonitor `search_web` outcomes during your task:\n- Count consecutive failures (empty results, 'unknown error', timeout, access denied)\n- Note the error type - distinguish between:\n  - **Recoverable**: Rate limiting, temporary timeout (retry 1-2 times)\n  - **Non-recoverable**: 'unknown error', persistent empty results, access denied\n\n### Step 2: Apply the 2-Failure Rule\n\n```\nIF search_web failures >= 2 (same or similar queries)\nTHEN:\n  1. Stop attempting search_web\n  2. Document what information you attempted to retrieve\n  3. Proceed with existing domain knowledge\n```\n\n### Step 3: Pivot to Domain Knowledge\n\nWhen pivoting:\n1. **Acknowledge the limitation**: Note that external verification was attempted but unavailable\n2. **Use established knowledge**: Draw on training data for well-known facts, laws, cases, standards\n3. **Generate with run_shell**: Create documents using Python\u002Fbash scripts rather than waiting for external data\n\nExample pivot workflow:\n```python\n# Instead of continuing search_web attempts:\n# 1. Compile known information from domain knowledge\nknown_facts = {\n    \"law\": \"COPPA requirements for children's data\",\n    \"precedent\": \"FTC v. Google\u002FYouTube settlement patterns\",\n    \"jurisdiction\": \"California privacy law framework\"\n}\n\n# 2. Generate document directly\n# Use run_shell with Python to create PDF\u002FDOC\n```\n\n### Step 4: Document Generation Pattern\n\nWhen generating documents after pivoting:\n\n```bash\n# Use run_shell to execute document generation\n# Example: Create legal memo with Python + reportlab\npython \u003C\u003C 'EOF'\nfrom reportlab.lib.pagesizes import letter\nfrom reportlab.pdfgen import canvas\n\n# Generate document with domain knowledge\nc = canvas.Canvas(\"memo.pdf\", pagesize=letter)\nc.drawString(100, 750, \"Legal Analysis based on established precedents\")\n# ... continue with known information\nc.save()\nEOF\n```\n\n## When This Applies\n\nThis heuristic is particularly useful for:\n\n| Task Type | Why It Applies |\n|-----------|----------------|\n| Legal research | Case law and statutes are well-documented in training |\n| Technical documentation | Standards and best practices are established knowledge |\n| Historical analysis | Past events and data are in training corpus |\n| Regulatory compliance | Major regulations (GDPR, COPPA, HIPAA) are well-known |\n\n## When NOT to Pivot\n\nDo NOT apply this heuristic when:\n\n- You need **current\u002Ftime-sensitive data** (stock prices, weather, news \u003C 24hrs)\n- The task **explicitly requires** external verification\n- You have **not yet attempted** the search (first attempt should complete)\n- Errors are clearly **recoverable** (rate limit with retry-after header)\n\n## Example Decision Flow\n\n```\nAttempt search_web → Success? → Yes → Use results\n                         ↓\n                        No\n                         ↓\n            Is this failure #1? → Yes → Retry once with modified query\n                         ↓\n                        No (failure #2+)\n                         ↓\n            Pivot to domain knowledge + run_shell document generation\n```\n\n## Benefits\n\n- **Saves iterations**: Avoids wasting 3-4 attempts on unavailable data\n- **Maintains progress**: Task continues instead of stalling on search\n- **Leverages training**: Uses the substantial knowledge already available\n- **Clear handoff**: Explicitly documents the pivot decision for transparency\n\n## Anti-Patterns to Avoid\n\n❌ Continuing search_web attempts beyond 2 failures  \n❌ Not documenting why you pivoted from search  \n❌ Applying this to tasks requiring real-time data  \n❌ Abandoning search on first failure without retry  \n\n✅ Apply 2-failure rule consistently  \n✅ Document the pivot decision in your output  \n✅ Use run_shell efficiently after pivoting  \n✅ Distinguish recoverable vs non-recoverable errors","# Search Tool Failure Detection and Pivot Strategy\n\n## Purpose\n\nWhen using `search_web` or similar external data retrieval tools, repeated failures waste iterations. This skill provides a diagnostic heuristic to recognize when to abandon search attempts and proceed with existing domain knowledge.\n\n## Core Heuristic\n\n**If `search_web` fails 2+ consecutive times with 'unknown error' or similar non-recoverable errors, pivot to domain knowledge + `run_shell` for document generation.**\n\nDo not attempt more than 2 retries on the same or similar queries before pivoting.\n\n## Step-by-Step Instructions\n\n### Step 1: Track Search Failures\n\nMonitor `search_web` outcomes during your task:\n- Count consecutive failures (empty results, 'unknown error', timeout, access denied)\n- Note the error type - distinguish between:\n  - **Recoverable**: Rate limiting, temporary timeout (retry 1-2 times)\n  - **Non-recoverable**: 'unknown error', persistent empty results, access denied\n\n### Step 2: Apply the 2-Failure Rule\n\n```\nIF search_web failures >= 2 (same or similar queries)\nTHEN:\n  1. Stop attempting search_web\n  2. Document what information you attempted to retrieve\n  3. Proceed with existing domain knowledge\n```\n\n### Step 3: Pivot to Domain Knowledge\n\nWhen pivoting:\n1. **Acknowledge the limitation**: Note that external verification was attempted but unavailable\n2. **Use established knowledge**: Draw on training data for well-known facts, laws, cases, standards\n3. **Generate with run_shell**: Create documents using Python\u002Fbash scripts rather than waiting for external data\n\nExample pivot workflow:\n```python\n# Instead of continuing search_web attempts:\n# 1. Compile known information from domain knowledge\nknown_facts = {\n    \"law\": \"COPPA requirements for children's data\",\n    \"precedent\": \"FTC v. Google\u002FYouTube settlement patterns\",\n    \"jurisdiction\": \"California privacy law framework\"\n}\n\n# 2. Generate document directly\n# Use run_shell with Python to create PDF\u002FDOC\n```\n\n### Step 4: Document Generation Pattern\n\nWhen generating documents after pivoting:\n\n```bash\n# Use run_shell to execute document generation\n# Example: Create legal memo with Python + reportlab\npython \u003C\u003C 'EOF'\nfrom reportlab.lib.pagesizes import letter\nfrom reportlab.pdfgen import canvas\n\n# Generate document with domain knowledge\nc = canvas.Canvas(\"memo.pdf\", pagesize=letter)\nc.drawString(100, 750, \"Legal Analysis based on established precedents\")\n# ... continue with known information\nc.save()\nEOF\n```\n\n## When This Applies\n\nThis heuristic is particularly useful for:\n\n| Task Type | Why It Applies |\n|-----------|----------------|\n| Legal research | Case law and statutes are well-documented in training |\n| Technical documentation | Standards and best practices are established knowledge |\n| Historical analysis | Past events and data are in training corpus |\n| Regulatory compliance | Major regulations (GDPR, COPPA, HIPAA) are well-known |\n\n## When NOT to Pivot\n\nDo NOT apply this heuristic when:\n\n- You need **current\u002Ftime-sensitive data** (stock prices, weather, news \u003C 24hrs)\n- The task **explicitly requires** external verification\n- You have **not yet attempted** the search (first attempt should complete)\n- Errors are clearly **recoverable** (rate limit with retry-after header)\n\n## Example Decision Flow\n\n```\nAttempt search_web → Success? → Yes → Use results\n                         ↓\n                        No\n                         ↓\n            Is this failure #1? → Yes → Retry once with modified query\n                         ↓\n                        No (failure #2+)\n                         ↓\n            Pivot to domain knowledge + run_shell document generation\n```\n\n## Benefits\n\n- **Saves iterations**: Avoids wasting 3-4 attempts on unavailable data\n- **Maintains progress**: Task continues instead of stalling on search\n- **Leverages training**: Uses the substantial knowledge already available\n- **Clear handoff**: Explicitly documents the pivot decision for transparency\n\n## Anti-Patterns to Avoid\n\n❌ Continuing search_web attempts beyond 2 failures  \n❌ Not documenting why you pivoted from search  \n❌ Applying this to tasks requiring real-time data  \n❌ Abandoning search on first failure without retry  \n\n✅ Apply 2-failure rule consistently  \n✅ Document the pivot decision in your output  \n✅ Use run_shell efficiently after pivoting  \n✅ Distinguish recoverable vs non-recoverable errors",{"name":5,"description":7},"main","6b84b0689ca0d9e7b1e4061188ff6b60758e0f8a","openspace\u002Flegacy\u002Fv1\u002Fgdpval_bench\u002Fskills\u002Fsearch-fail-pivot",944,"https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot","https:\u002F\u002Fgithub.com\u002FHKUDS","https:\u002F\u002Fsites.google.com\u002Fview\u002Fchaoh","2026-11-12 22:26:39",{"source":8,"skillName":5,"commands":44,"prompt":54,"zipAvailable":55},[45,48,51],{"tool":46,"command":47},"npx","npx skills add HKUDS\u002FAgentPilot --skill search-fail-pivot",{"tool":49,"command":50},"bunx","bunx skills add HKUDS\u002FAgentPilot --skill search-fail-pivot",{"tool":52,"command":53},"pnpm","pnpm dlx skills add HKUDS\u002FAgentPilot --skill search-fail-pivot","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Flegacy\u002Fv1\u002Fgdpval_bench\u002Fskills\u002Fsearch-fail-pivot\u002FSKILL.md",true,[57],{"path":58,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":61,"sourceHtmlUrl":62},".skill_id","OTHER","其他",31,"https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Flegacy\u002Fv1\u002Fgdpval_bench\u002Fskills\u002Fsearch-fail-pivot\u002F.skill_id",[64,84,103,127,149,172],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":30},"01M6HYB297NDABRERJRVG0K8R0","agentpilot","Set up AgentPilot and use its single entry to prepare missing inputs and run configured RSI experiments. Use when the user requests AgentPilot installation or work with AgentPilot.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fplugins\u002Fagentpilot\u002Fskills\u002Fagentpilot\u002FSKILL.md",[],[66,71,72,73,74,75,76,77,78,79,80,81,82,83],"prepare","requests","with","missing","installation","rsi","up","use","user","work","SET","run","experiments",{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":30},"01M6HYAYWXTTE1J7DJM0MFNQ4H","remember","Review auto-memory entries and propose promotions to OPENSPACE.md, OPENSPACE.local.md, or retained auto-memory topics; detect duplicates, stale facts, conflicts, and ambiguous entries before making changes.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fopenspace\u002Fskills\u002Fremember\u002FSKILL.md",[],[86,91,92,93,94,95,96,97,98,99,100,101,102],"memory","review","auto","openspace","changes","facts","detect","entries","local","conflicts","stale","duplicates",{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":114,"publishedAt":30},"01M6HYAVMPM4SBDV60R2B5HPV8","skill-discovery","Search reusable OpenSpace skills locally, and browse cloud packages step by step when a cloud skill may help. Reusing proven skills saves tokens, improves reliability, and extends your capabilities beyond built-in tools.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fopenspace\u002Fhost_skills\u002Fskill-discovery\u002FSKILL.md",[109],{"categoryNo":110,"categoryName":111,"taxonomyType":112,"taxonomyTypeDesc":113},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[115,116,117,94,118,119,120,121,122,123,124,125,19,126],"skill discovery","discovery","reusable","step by step","tokens","packages","help","cloud","reliability","capabilities","browse","Tools",{"skillNo":128,"name":129,"nameZh":6,"description":130,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":131,"categories":132,"keywords":134,"publishedAt":30},"01M6HYAR811B3RB0M50RBEDGN6","delegate-task","Delegate tasks to OpenSpace — a full-stack autonomous worker for coding, DevOps, web research, and desktop automation, backed by an extensive MCP tool and skill library. Skills auto-improve through use, reducing token consumption over time. A cloud community lets agents share and collectively evolve reusable skills.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Fopenspace\u002Fhost_skills\u002Fdelegate-task\u002FSKILL.md",[133],{"categoryNo":110,"categoryName":111,"taxonomyType":112,"taxonomyTypeDesc":113},[135,136,137,94,138,139,140,141,142,117,93,78,143,23,144,29,145,146,147,148],"delegate","task","delegate task","time","improve","desktop automation","worker","community","autonomous","stack","share","tasks","automation","MCP",{"skillNo":150,"name":151,"nameZh":6,"description":152,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":153,"categories":154,"keywords":155,"publishedAt":30},"01M6HYAM9FSGV6E52AYENDQXYH","write-special-chars-ts-check","Workaround for write_file failures caused by special characters (apostrophes, backticks, template literals) using a Python heredoc script, plus a reliable TypeScript type-checking step that filters output to only the newly-written files to avoid being misled by pre-existing errors.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Flegacy\u002Fv1\u002Fshowcase\u002Fskills\u002Fwrite-special-chars-ts-check\u002FSKILL.md",[],[156,157,158,159,160,161,162,163,164,165,166,167,168,169,29,170,171],"write special chars ts check","write","check","type","files","filters","only","template","file","python","typescript","output","Existing","type-checking","errors","script",{"skillNo":173,"name":174,"nameZh":6,"description":175,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":176,"categories":177,"keywords":178,"publishedAt":30},"01M6HYAH05HGV7SVWSVR8PJRQD","write-special-chars-readback","Workaround for write_file failures caused by special characters (apostrophes, backticks, template literals) using a Python heredoc script, with mandatory post-write verification via read_file and cleanup of \u002Ftmp helper scripts.","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FAgentPilot\u002Fblob\u002Fmain\u002Fopenspace\u002Flegacy\u002Fv1\u002Fshowcase\u002Fskills\u002Fwrite-special-chars-readback\u002FSKILL.md",[],[179,180,157,73,29,171,181,182,183,184,185,163,164,186,187,165],"write special chars readback","readback","helper","post","via","scripts","read","verification","cleanup"]