[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6DB0S125VVD0ARSH03THERC":3,"skill-same:01M6DB0S125VVD0ARSH03THERC":61},{"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":35,"linkOnly":36,"markdownContent":37,"markdownBody":38,"frontMatter":39,"refName":41,"commitSha":42,"skillDirectory":43,"forksCount":14,"repositoryHtmlUrl":44,"repositoryHomepageUrl":6,"ownerHtmlUrl":45,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":46,"install":47,"resources":60},"01M6DB0S125VVD0ARSH03THERC","agent-introspection-debugging",null,"Structured self-debugging workflow for AI agent failures using capture, diagnosis, contained recovery, and introspection reports.","Prekzursil\u002Fagent-skills-toolchain","Prekzursil","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F54636077?v=4","01M6DACZ1PCQHD99SCR1YXG28B","Andrei Mihai Visalon","01M6DACZ670WZF535V3WYYSS4Z",0,"MIT","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fagent-introspection-debugging\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],"debugging","introspection","agent introspection debugging","ai agent","workflow","recovery","capture","reports","self","structured","diagnosis","2026-11-10 10:12:30",false,"---\nname: agent-introspection-debugging\ndescription: Structured self-debugging workflow for AI agent failures using capture, diagnosis, contained recovery, and introspection reports.\norigin: ECC\n---\n\n# Agent Introspection Debugging\n\nUse this skill when an agent run is failing repeatedly, consuming tokens without progress, looping on the same tools, or drifting away from the intended task.\n\nThis is a workflow skill, not a hidden runtime. It teaches the agent to debug itself systematically before escalating to a human.\n\n## When to Activate\n\n- Maximum tool call \u002F loop-limit failures\n- Repeated retries with no forward progress\n- Context growth or prompt drift that starts degrading output quality\n- File-system or environment state mismatch between expectation and reality\n- Tool failures that are likely recoverable with diagnosis and a smaller corrective action\n\n## Scope Boundaries\n\nActivate this skill for:\n- capturing failure state before retrying blindly\n- diagnosing common agent-specific failure patterns\n- applying contained recovery actions\n- producing a structured human-readable debug report\n\nDo not use this skill as the primary source for:\n- feature verification after code changes; use `verification-loop`\n- framework-specific debugging when a narrower ECC skill already exists\n- runtime promises the current harness cannot enforce automatically\n\n## Four-Phase Loop\n\n### Phase 1: Failure Capture\n\nBefore trying to recover, record the failure precisely.\n\nCapture:\n- error type, message, and stack trace when available\n- last meaningful tool call sequence\n- what the agent was trying to do\n- current context pressure: repeated prompts, oversized pasted logs, duplicated plans, or runaway notes\n- current environment assumptions: cwd, branch, relevant service state, expected files\n\nMinimum capture template:\n\n```markdown\n## Failure Capture\n- Session \u002F task:\n- Goal in progress:\n- Error:\n- Last successful step:\n- Last failed tool \u002F command:\n- Repeated pattern seen:\n- Environment assumptions to verify:\n```\n\n### Phase 2: Root-Cause Diagnosis\n\nMatch the failure to a known pattern before changing anything.\n\n| Pattern | Likely Cause | Check |\n| --- | --- | --- |\n| Maximum tool calls \u002F repeated same command | loop or no-exit observer path | inspect the last N tool calls for repetition |\n| Context overflow \u002F degraded reasoning | unbounded notes, repeated plans, oversized logs | inspect recent context for duplication and low-signal bulk |\n| `ECONNREFUSED` \u002F timeout | service unavailable or wrong port | verify service health, URL, and port assumptions |\n| `429` \u002F quota exhaustion | retry storm or missing backoff | count repeated calls and inspect retry spacing |\n| file missing after write \u002F stale diff | race, wrong cwd, or branch drift | re-check path, cwd, git status, and actual file existence |\n| tests still failing after “fix” | wrong hypothesis | isolate the exact failing test and re-derive the bug |\n\nDiagnosis questions:\n- is this a logic failure, state failure, environment failure, or policy failure?\n- did the agent lose the real objective and start optimizing the wrong subtask?\n- is the failure deterministic or transient?\n- what is the smallest reversible action that would validate the diagnosis?\n\n### Phase 3: Contained Recovery\n\nRecover with the smallest action that changes the diagnosis surface.\n\nSafe recovery actions:\n- stop repeated retries and restate the hypothesis\n- trim low-signal context and keep only the active goal, blockers, and evidence\n- re-check the actual filesystem \u002F branch \u002F process state\n- narrow the task to one failing command, one file, or one test\n- switch from speculative reasoning to direct observation\n- escalate to a human when the failure is high-risk or externally blocked\n\nDo not claim unsupported auto-healing actions like “reset agent state” or “update harness config” unless you are actually doing them through real tools in the current environment.\n\nContained recovery checklist:\n\n```markdown\n## Recovery Action\n- Diagnosis chosen:\n- Smallest action taken:\n- Why this is safe:\n- What evidence would prove the fix worked:\n```\n\n### Phase 4: Introspection Report\n\nEnd with a report that makes the recovery legible to the next agent or human.\n\n```markdown\n## Agent Self-Debug Report\n- Session \u002F task:\n- Failure:\n- Root cause:\n- Recovery action:\n- Result: success | partial | blocked\n- Token \u002F time burn risk:\n- Follow-up needed:\n- Preventive change to encode later:\n```\n\n## Recovery Heuristics\n\nPrefer these interventions in order:\n\n1. Restate the real objective in one sentence.\n2. Verify the world state instead of trusting memory.\n3. Shrink the failing scope.\n4. Run one discriminating check.\n5. Only then retry.\n\nBad pattern:\n- retrying the same action three times with slightly different wording\n\nGood pattern:\n- capture failure\n- classify the pattern\n- run one direct check\n- change the plan only if the check supports it\n\n## Integration with ECC\n\n- Use `verification-loop` after recovery if code was changed.\n- Use `continuous-learning-v2` when the failure pattern is worth turning into an instinct or later skill.\n- Use `council` when the issue is not technical failure but decision ambiguity.\n- Use `workspace-surface-audit` if the failure came from conflicting local state or repo drift.\n\n## Output Standard\n\nWhen this skill is active, do not end with “I fixed it” alone.\n\nAlways provide:\n- the failure pattern\n- the root-cause hypothesis\n- the recovery action\n- the evidence that the situation is now better or still blocked\n","# Agent Introspection Debugging\n\nUse this skill when an agent run is failing repeatedly, consuming tokens without progress, looping on the same tools, or drifting away from the intended task.\n\nThis is a workflow skill, not a hidden runtime. It teaches the agent to debug itself systematically before escalating to a human.\n\n## When to Activate\n\n- Maximum tool call \u002F loop-limit failures\n- Repeated retries with no forward progress\n- Context growth or prompt drift that starts degrading output quality\n- File-system or environment state mismatch between expectation and reality\n- Tool failures that are likely recoverable with diagnosis and a smaller corrective action\n\n## Scope Boundaries\n\nActivate this skill for:\n- capturing failure state before retrying blindly\n- diagnosing common agent-specific failure patterns\n- applying contained recovery actions\n- producing a structured human-readable debug report\n\nDo not use this skill as the primary source for:\n- feature verification after code changes; use `verification-loop`\n- framework-specific debugging when a narrower ECC skill already exists\n- runtime promises the current harness cannot enforce automatically\n\n## Four-Phase Loop\n\n### Phase 1: Failure Capture\n\nBefore trying to recover, record the failure precisely.\n\nCapture:\n- error type, message, and stack trace when available\n- last meaningful tool call sequence\n- what the agent was trying to do\n- current context pressure: repeated prompts, oversized pasted logs, duplicated plans, or runaway notes\n- current environment assumptions: cwd, branch, relevant service state, expected files\n\nMinimum capture template:\n\n```markdown\n## Failure Capture\n- Session \u002F task:\n- Goal in progress:\n- Error:\n- Last successful step:\n- Last failed tool \u002F command:\n- Repeated pattern seen:\n- Environment assumptions to verify:\n```\n\n### Phase 2: Root-Cause Diagnosis\n\nMatch the failure to a known pattern before changing anything.\n\n| Pattern | Likely Cause | Check |\n| --- | --- | --- |\n| Maximum tool calls \u002F repeated same command | loop or no-exit observer path | inspect the last N tool calls for repetition |\n| Context overflow \u002F degraded reasoning | unbounded notes, repeated plans, oversized logs | inspect recent context for duplication and low-signal bulk |\n| `ECONNREFUSED` \u002F timeout | service unavailable or wrong port | verify service health, URL, and port assumptions |\n| `429` \u002F quota exhaustion | retry storm or missing backoff | count repeated calls and inspect retry spacing |\n| file missing after write \u002F stale diff | race, wrong cwd, or branch drift | re-check path, cwd, git status, and actual file existence |\n| tests still failing after “fix” | wrong hypothesis | isolate the exact failing test and re-derive the bug |\n\nDiagnosis questions:\n- is this a logic failure, state failure, environment failure, or policy failure?\n- did the agent lose the real objective and start optimizing the wrong subtask?\n- is the failure deterministic or transient?\n- what is the smallest reversible action that would validate the diagnosis?\n\n### Phase 3: Contained Recovery\n\nRecover with the smallest action that changes the diagnosis surface.\n\nSafe recovery actions:\n- stop repeated retries and restate the hypothesis\n- trim low-signal context and keep only the active goal, blockers, and evidence\n- re-check the actual filesystem \u002F branch \u002F process state\n- narrow the task to one failing command, one file, or one test\n- switch from speculative reasoning to direct observation\n- escalate to a human when the failure is high-risk or externally blocked\n\nDo not claim unsupported auto-healing actions like “reset agent state” or “update harness config” unless you are actually doing them through real tools in the current environment.\n\nContained recovery checklist:\n\n```markdown\n## Recovery Action\n- Diagnosis chosen:\n- Smallest action taken:\n- Why this is safe:\n- What evidence would prove the fix worked:\n```\n\n### Phase 4: Introspection Report\n\nEnd with a report that makes the recovery legible to the next agent or human.\n\n```markdown\n## Agent Self-Debug Report\n- Session \u002F task:\n- Failure:\n- Root cause:\n- Recovery action:\n- Result: success | partial | blocked\n- Token \u002F time burn risk:\n- Follow-up needed:\n- Preventive change to encode later:\n```\n\n## Recovery Heuristics\n\nPrefer these interventions in order:\n\n1. Restate the real objective in one sentence.\n2. Verify the world state instead of trusting memory.\n3. Shrink the failing scope.\n4. Run one discriminating check.\n5. Only then retry.\n\nBad pattern:\n- retrying the same action three times with slightly different wording\n\nGood pattern:\n- capture failure\n- classify the pattern\n- run one direct check\n- change the plan only if the check supports it\n\n## Integration with ECC\n\n- Use `verification-loop` after recovery if code was changed.\n- Use `continuous-learning-v2` when the failure pattern is worth turning into an instinct or later skill.\n- Use `council` when the issue is not technical failure but decision ambiguity.\n- Use `workspace-surface-audit` if the failure came from conflicting local state or repo drift.\n\n## Output Standard\n\nWhen this skill is active, do not end with “I fixed it” alone.\n\nAlways provide:\n- the failure pattern\n- the root-cause hypothesis\n- the recovery action\n- the evidence that the situation is now better or still blocked\n",{"name":5,"origin":40,"description":7},"ECC","main","a298c929b6cbd662e1fce54302ce520d5ceea1a5",".agents\u002Fskills\u002Fagent-introspection-debugging","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain","https:\u002F\u002Fgithub.com\u002FPrekzursil","2026-11-02 14:30:57",{"source":8,"skillName":5,"commands":48,"prompt":58,"zipAvailable":59},[49,52,55],{"tool":50,"command":51},"npx","npx skills add Prekzursil\u002Fagent-skills-toolchain --skill agent-introspection-debugging",{"tool":53,"command":54},"bunx","bunx skills add Prekzursil\u002Fagent-skills-toolchain --skill agent-introspection-debugging",{"tool":56,"command":57},"pnpm","pnpm dlx skills add Prekzursil\u002Fagent-skills-toolchain --skill agent-introspection-debugging","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fagent-introspection-debugging\u002FSKILL.md",true,[],[62,84,110,137,151,167],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":35},"01M6DF7D7478R55P46REYH4QAC","clarify-requirements","Resolve genuine pre-execution ambiguity by scanning repo context, generating 1-6 grounded options, and asking 1-3 focused questions via AskUserQuestion. NOT for mid-task use (injection risk); NOT when the goal is already unambiguous; use to disambiguate vague prompts before any file edits.","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Fworkspace-hooks\u002Fskills\u002Fclarify-requirements\u002FSKILL.md",[],[69,70,71,72,73,74,75,76,77,78,79,80,81,82,83],"clarify requirements","clarify","requirements","risk","options","task","prompts","goal","scanning","use","context","repo","injection","execution","file",{"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":91,"publishedAt":35},"01M6DF79CA4BHJS66J8EZDBPWJ","openmm-md","Run FREE, open-source, Windows-native GPU molecular dynamics via the openmm-md MCP (OpenMM + CUDA on the local NVIDIA GPU). Use for protein \u002F protein-ligand MD when you want a free stack or are on Windows without a Schrodinger license. Covers the full workflow (prepare, solvate, minimize, equilibrate, production, analyze, mmgbsa) AND an honest decision guide vs Schrodinger Desmond, including how to run Desmond legitimately on Linux or a cloud GPU with a real license. This is a deliberately LESSER free alternative and does NOT replace the schrodinger MCP tools. Do NOT use it to pick a free tool when a validated FEP+, WaterMap, IFD-MD, or Prime MM-GBSA result is actually required.","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Fmcp\u002Fopenmm-md-mcp\u002Fskill\u002Fopenmm-md\u002FSKILL.md",[90],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[92,93,94,95,96,97,98,99,100,101,102,103,28,104,78,105,106,107,108,109],"openmm md","how","decision","pick","linux","analyze","tool","stack","open","NVIDIA","gpu","license","windows","free","molecular dynamics","local","do","MCP",{"skillNo":111,"name":112,"nameZh":6,"description":113,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":114,"categories":115,"keywords":122,"publishedAt":35},"01M6DF76C1MG85VEWP8VZCBNMN","visual-review","Take screenshots of web pages and UI using Playwright for visual review and iteration","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Feditor\u002Fmetaswarm\u002Fskills\u002Fvisual-review\u002FSKILL.md",[116,119],{"categoryNo":117,"categoryName":118,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000003","演示文稿",{"categoryNo":120,"categoryName":121,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000008","测试与质量",[123,124,125,126,127,128,129,130,131,132,93,133,134,135,136],"review","visual","visual review","pages","web","false","iteration","take","check","slides","ui","screenshot","playwright","screenshots",{"skillNo":138,"name":139,"nameZh":6,"description":140,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":141,"categories":142,"keywords":143,"publishedAt":35},"01M6DF73AQVCBZKEN0YXEX9FXP","status","Diagnostic status report — shows metaswarm installation state, project setup, and potential issues","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Feditor\u002Fmetaswarm\u002Fskills\u002Fstatus\u002FSKILL.md",[],[139,144,145,146,147,148,149,150],"setup","report","project","installation","project setup","issues","state",{"skillNo":152,"name":153,"nameZh":6,"description":154,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":155,"categories":156,"keywords":158,"publishedAt":35},"01M6DF6X6C2XB0P6GX76P7TTYV","start","Use when starting work on any task, when the user mentions metaswarm, or when the user wants to begin tracked development work","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Feditor\u002Fmetaswarm\u002Fskills\u002Fstart\u002FSKILL.md",[157],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[153,159,74,160,161,162,78,163,164,165,166],"start issue","issue","ready","work on issue","user","work","work on","development",{"skillNo":168,"name":144,"nameZh":6,"description":169,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":170,"categories":171,"keywords":172,"publishedAt":35},"01M6DF6QMJ9AJKFKHHKRT005PQ","Interactive project setup — detects your project, configures metaswarm, writes project-local files","https:\u002F\u002Fgithub.com\u002FPrekzursil\u002Fagent-skills-toolchain\u002Fblob\u002Fmain\u002Ftools\u002Feditor\u002Fmetaswarm\u002Fskills\u002Fsetup\u002FSKILL.md",[],[144,146,107,173,148,174],"files","interactive"]