{
  "schemaVersion": "1.0",
  "item": {
    "slug": "input-classification-v1",
    "name": "input classification",
    "source": "tencent",
    "type": "skill",
    "category": "AI 智能",
    "sourceUrl": "https://clawhub.ai/omprasad122007-rgb/input-classification-v1",
    "canonicalUrl": "https://clawhub.ai/omprasad122007-rgb/input-classification-v1",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadMode": "redirect",
    "downloadUrl": "/downloads/input-classification-v1",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=input-classification-v1",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "installMethod": "Manual import",
    "extraction": "Extract archive",
    "prerequisites": [
      "OpenClaw"
    ],
    "packageFormat": "ZIP package",
    "includedAssets": [
      "SKILL.md",
      "references/classification-models.md",
      "references/system-integration.md"
    ],
    "primaryDoc": "SKILL.md",
    "quickSetup": [
      "Download the package from Yavira.",
      "Extract the archive and review SKILL.md first.",
      "Import or place the package into your OpenClaw setup."
    ],
    "agentAssist": {
      "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.",
      "steps": [
        "Download the package from Yavira.",
        "Extract it into a folder your agent can access.",
        "Paste one of the prompts below and point your agent at the extracted folder."
      ],
      "prompts": [
        {
          "label": "New install",
          "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete."
        },
        {
          "label": "Upgrade existing",
          "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run."
        }
      ]
    },
    "sourceHealth": {
      "source": "tencent",
      "slug": "input-classification-v1",
      "status": "healthy",
      "reason": "direct_download_ok",
      "recommendedAction": "download",
      "checkedAt": "2026-05-04T23:06:49.677Z",
      "expiresAt": "2026-05-11T23:06:49.677Z",
      "httpStatus": 200,
      "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=input-classification-v1",
      "contentType": "application/zip",
      "probeMethod": "head",
      "details": {
        "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=input-classification-v1",
        "contentDisposition": "attachment; filename=\"input-classification-v1-1.0.0.zip\"",
        "redirectLocation": null,
        "bodySnippet": null,
        "slug": "input-classification-v1"
      },
      "scope": "item",
      "summary": "Item download looks usable.",
      "detail": "Yavira can redirect you to the upstream package for this item.",
      "primaryActionLabel": "Download for OpenClaw",
      "primaryActionHref": "/downloads/input-classification-v1"
    },
    "validation": {
      "installChecklist": [
        "Use the Yavira download entry.",
        "Review SKILL.md after the package is downloaded.",
        "Confirm the extracted package contains the expected setup assets."
      ],
      "postInstallChecks": [
        "Confirm the extracted package includes the expected docs or setup files.",
        "Validate the skill or prompts are available in your target agent workspace.",
        "Capture any manual follow-up steps the agent could not complete."
      ]
    },
    "downloadPageUrl": "https://openagent3.xyz/downloads/input-classification-v1",
    "agentPageUrl": "https://openagent3.xyz/skills/input-classification-v1/agent",
    "manifestUrl": "https://openagent3.xyz/skills/input-classification-v1/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/input-classification-v1/agent.md"
  },
  "agentAssist": {
    "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.",
    "steps": [
      "Download the package from Yavira.",
      "Extract it into a folder your agent can access.",
      "Paste one of the prompts below and point your agent at the extracted folder."
    ],
    "prompts": [
      {
        "label": "New install",
        "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete."
      },
      {
        "label": "Upgrade existing",
        "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run."
      }
    ]
  },
  "documentation": {
    "source": "clawhub",
    "primaryDoc": "SKILL.md",
    "sections": [
      {
        "title": "1. Skill Name",
        "body": "Input Classification System\n\nIdentifier: input-classification-system"
      },
      {
        "title": "2. Version",
        "body": "1.0\n\nThis is the initial release of the Input Classification System skill."
      },
      {
        "title": "3. Skill Purpose",
        "body": "Provide a deterministic, rule-based classification system that enables the Main Agent to categorize clarified user input into exactly one primary task category, assign execution complexity, assess risk level, and determine confidence score before any task decomposition or execution planning occurs.\n\nMeasurable Objectives:\n\nAchieve 100% single-category classification (no ambiguous multi-category outputs)\nProvide deterministic tie-breaking for all edge cases\nAssign complexity levels using measurable thresholds\nCalculate confidence scores with explicit formulas\nRoute ambiguous inputs to clarification with specific triggers"
      },
      {
        "title": "4. What This Skill Does",
        "body": "Classifies clarified input into exactly one primary category from a fixed list\nAssigns complexity level based on measurable operation counts and time estimates\nCalculates risk level based on impact scope and reversibility criteria\nComputes confidence score using explicit scoring factors\nDetermines if input requires additional clarification\nSets the appropriate state transition after classification\nApplies deterministic tie-breaking logic when multiple categories match\nValidates that input has been clarified before classification\nLogs all classification decisions with full context\nProvides secondary tags for additional context (maximum 3)\nEnforces boundary definitions between categories\nTriggers escalation for high-risk or low-confidence classifications\nPrevents misclassification through explicit rules\nOutputs structured ClassificationResult for downstream systems\nMaintains audit trail for all classification decisions"
      },
      {
        "title": "5. What This Skill Must Not Do",
        "body": "Must not perform task decomposition or breakdown\nMust not solve or execute any tasks\nMust not create execution plans or strategies\nMust not apply emotional reasoning or sentiment analysis\nMust not assign multiple primary categories to a single input\nMust not overlap with Clarification System responsibilities\nMust not modify the original input content\nMust not make assumptions about user intent without explicit indicators\nMust not skip complexity assessment for any classification\nMust not bypass risk assessment for any classification\nMust not assign confidence scores below threshold without escalation\nMust not classify inputs that have not been clarified first\nMust not use probabilistic or non-deterministic classification methods\nMust not ignore tie-breaking rules when categories conflict\nMust not proceed to execution without setting state transition"
      },
      {
        "title": "6. Activation Conditions",
        "body": "This skill activates when ALL of the following conditions are met:\n\nClarification Complete: Input has been processed by the Clarification System and marked as clarified\nNo Pending Questions: No outstanding clarification questions remain\nClassification Not Yet Performed: Input has not been previously classified\nValid Input Structure: Input contains recognizable task indicators\n\nPre-Activation Checklist:\n\nInput marked as \"clarified\" by Clarification System\n No clarification questions pending\n Input contains actionable request\n Input is not empty or malformed\n\nDo NOT activate if:\n\nInput is still being clarified\nInput is a clarification question itself\nInput is a response to a clarification question\nInput has already been classified"
      },
      {
        "title": "7. Classification Categories",
        "body": "The following 15 categories form the fixed classification list:\n\nCategoryCodeDescriptionCODE_GENERATIONCGWriting, modifying, or generating source codeCODE_REVIEWCRReviewing, analyzing, or auditing existing codeDEBUGGINGDBIdentifying and fixing bugs, errors, or issuesDATA_ANALYSISDAAnalyzing, processing, or visualizing dataFILE_OPERATIONSFOReading, writing, moving, or managing filesDOCUMENTATIONDCCreating or updating documentationREFACTORINGRFRestructuring code without changing behaviorTESTINGTSWriting, running, or managing testsDEPLOYMENTDPDeploying applications or infrastructureRESEARCHRSInvestigating, searching, or gathering informationCONFIGURATIONCFSetting up or modifying configurationsCOMMUNICATIONCMDrafting messages, emails, or communicationsCONVERSIONCVTransforming data or files between formatsANALYSISANGeneral analysis not covered by other categoriesPLANNINGPLCreating plans, strategies, or roadmaps\n\nCategory Priority Order (for tie-breaking):\n\nDEBUGGING (highest - immediate attention needed)\nDEPLOYMENT\nCODE_GENERATION\nREFACTORING\nTESTING\nCODE_REVIEW\nDATA_ANALYSIS\nCONFIGURATION\nFILE_OPERATIONS\nCONVERSION\nDOCUMENTATION\nRESEARCH\nANALYSIS\nCOMMUNICATION\nPLANNING (lowest)"
      },
      {
        "title": "CODE_GENERATION vs CODE_REVIEW",
        "body": "CODE_GENERATION: Input requests creating new code or modifying existing code\nCODE_REVIEW: Input requests analysis of existing code without modifications\nBoundary: If modification is implied, classify as CODE_GENERATION"
      },
      {
        "title": "CODE_GENERATION vs REFACTORING",
        "body": "CODE_GENERATION: Creating new functionality or features\nREFACTORING: Restructuring existing code without new functionality\nBoundary: \"Improve performance\" without feature change = REFACTORING"
      },
      {
        "title": "DEBUGGING vs CODE_GENERATION",
        "body": "DEBUGGING: Input explicitly mentions errors, bugs, or failures\nCODE_GENERATION: Input requests new features without error context\nBoundary: \"Fix this bug\" = DEBUGGING; \"Add error handling\" = CODE_GENERATION"
      },
      {
        "title": "DATA_ANALYSIS vs ANALYSIS",
        "body": "DATA_ANALYSIS: Input involves data processing, statistics, or visualization\nANALYSIS: Input involves general analysis of concepts, requirements, or situations\nBoundary: If data files/sets are mentioned = DATA_ANALYSIS"
      },
      {
        "title": "FILE_OPERATIONS vs CONVERSION",
        "body": "FILE_OPERATIONS: Input requests file management (read, write, move, delete)\nCONVERSION: Input requests format transformation between file types\nBoundary: \"Convert X to Y format\" = CONVERSION; \"Read file X\" = FILE_OPERATIONS"
      },
      {
        "title": "RESEARCH vs ANALYSIS",
        "body": "RESEARCH: Input requests information gathering or investigation\nANALYSIS: Input requests evaluation or assessment of known information\nBoundary: \"Find information about X\" = RESEARCH; \"Evaluate X\" = ANALYSIS"
      },
      {
        "title": "TESTING vs CODE_GENERATION",
        "body": "TESTING: Input specifically requests test creation or test execution\nCODE_GENERATION: Input requests production code\nBoundary: \"Write tests for X\" = TESTING; \"Write X with tests\" = CODE_GENERATION (primary)"
      },
      {
        "title": "CONFIGURATION vs DEPLOYMENT",
        "body": "CONFIGURATION: Input requests setup of settings or configurations\nDEPLOYMENT: Input requests deployment to environments or infrastructure\nBoundary: Local setup = CONFIGURATION; Remote/environment setup = DEPLOYMENT"
      },
      {
        "title": "9. Single-Primary-Category Rule",
        "body": "Rule: Every classified input MUST have exactly one primary category."
      },
      {
        "title": "Enforcement Rules",
        "body": "No Multi-Category Output: Never output multiple primary categories\nTie-Breaking Required: When multiple categories match, apply tie-breaking logic\nCategory Exclusivity: Primary category is mutually exclusive with other primary categories"
      },
      {
        "title": "Tie-Breaking Logic",
        "body": "When input matches multiple categories, apply in order:\n\nStep 1: Keyword Dominance\n\nCount explicit keyword matches for each candidate category\nCategory with highest keyword count wins\nIf tied, proceed to Step 2\n\nStep 2: Action Verb Analysis\n\nIdentify primary action verb in input\nMap action verb to category using verb-to-category mapping\nIf still tied, proceed to Step 3\n\nStep 3: Priority Order\n\nApply category priority order (see Section 7)\nHigher priority category wins\n\nStep 4: Default Fallback\n\nIf all steps fail, default to ANALYSIS category"
      },
      {
        "title": "Tie-Breaking Example",
        "body": "Input: \"Debug and fix the error in the authentication code, then add logging\"\n\nKeywords: DEBUGGING (debug, fix, error), CODE_GENERATION (add, logging)\nAction Verb: \"Debug\" (primary action) → DEBUGGING\nResult: DEBUGGING (primary), secondary_tags: [CODE_GENERATION]"
      },
      {
        "title": "10. Secondary Tag Rules",
        "body": "Secondary tags provide additional context without affecting primary routing."
      },
      {
        "title": "Rules",
        "body": "Maximum 3 Tags: No more than 3 secondary tags per classification\nNo Primary Duplicate: Secondary tags cannot duplicate primary category\nRelated Categories Only: Secondary tags must be from the fixed category list\nRelevance Threshold: Only add tags with >50% keyword match confidence"
      },
      {
        "title": "Secondary Tag Selection Process",
        "body": "Identify all categories with keyword matches (excluding primary)\nCalculate relevance score for each (0.0-1.0)\nSort by relevance score (descending)\nSelect top 3 with score > 0.5\nAdd to ClassificationResult"
      },
      {
        "title": "When to Apply Secondary Tags",
        "body": "Input contains multiple distinct sub-tasks\nInput references multiple technology domains\nInput implies follow-up work in other categories\nInput has context from previous interactions in different categories"
      },
      {
        "title": "When NOT to Apply Secondary Tags",
        "body": "Input is single-focused with no additional context\nRelevance scores are below threshold (0.5)\nWould duplicate primary category\nWould exceed 3-tag limit"
      },
      {
        "title": "Secondary Tag Examples",
        "body": "InputPrimarySecondary Tags\"Debug the API and update the docs\"DEBUGGING[DOCUMENTATION]\"Refactor the auth module and add tests\"REFACTORING[TESTING]\"Analyze sales data and create a report\"DATA_ANALYSIS[DOCUMENTATION, COMMUNICATION]\"Fix this bug\"DEBUGGING[]"
      },
      {
        "title": "Classification Models Reference",
        "body": "For detailed complexity, risk, confidence, and state transition models, see classification-models.md.\n\nFor system integration, failure conditions, logging, and examples, see system-integration.md."
      }
    ],
    "body": "Input Classification System\n1. Skill Name\n\nInput Classification System\n\nIdentifier: input-classification-system\n\n2. Version\n\n1.0\n\nThis is the initial release of the Input Classification System skill.\n\n3. Skill Purpose\n\nProvide a deterministic, rule-based classification system that enables the Main Agent to categorize clarified user input into exactly one primary task category, assign execution complexity, assess risk level, and determine confidence score before any task decomposition or execution planning occurs.\n\nMeasurable Objectives:\n\nAchieve 100% single-category classification (no ambiguous multi-category outputs)\nProvide deterministic tie-breaking for all edge cases\nAssign complexity levels using measurable thresholds\nCalculate confidence scores with explicit formulas\nRoute ambiguous inputs to clarification with specific triggers\n4. What This Skill Does\nClassifies clarified input into exactly one primary category from a fixed list\nAssigns complexity level based on measurable operation counts and time estimates\nCalculates risk level based on impact scope and reversibility criteria\nComputes confidence score using explicit scoring factors\nDetermines if input requires additional clarification\nSets the appropriate state transition after classification\nApplies deterministic tie-breaking logic when multiple categories match\nValidates that input has been clarified before classification\nLogs all classification decisions with full context\nProvides secondary tags for additional context (maximum 3)\nEnforces boundary definitions between categories\nTriggers escalation for high-risk or low-confidence classifications\nPrevents misclassification through explicit rules\nOutputs structured ClassificationResult for downstream systems\nMaintains audit trail for all classification decisions\n5. What This Skill Must Not Do\nMust not perform task decomposition or breakdown\nMust not solve or execute any tasks\nMust not create execution plans or strategies\nMust not apply emotional reasoning or sentiment analysis\nMust not assign multiple primary categories to a single input\nMust not overlap with Clarification System responsibilities\nMust not modify the original input content\nMust not make assumptions about user intent without explicit indicators\nMust not skip complexity assessment for any classification\nMust not bypass risk assessment for any classification\nMust not assign confidence scores below threshold without escalation\nMust not classify inputs that have not been clarified first\nMust not use probabilistic or non-deterministic classification methods\nMust not ignore tie-breaking rules when categories conflict\nMust not proceed to execution without setting state transition\n6. Activation Conditions\n\nThis skill activates when ALL of the following conditions are met:\n\nClarification Complete: Input has been processed by the Clarification System and marked as clarified\nNo Pending Questions: No outstanding clarification questions remain\nClassification Not Yet Performed: Input has not been previously classified\nValid Input Structure: Input contains recognizable task indicators\n\nPre-Activation Checklist:\n\n Input marked as \"clarified\" by Clarification System\n No clarification questions pending\n Input contains actionable request\n Input is not empty or malformed\n\nDo NOT activate if:\n\nInput is still being clarified\nInput is a clarification question itself\nInput is a response to a clarification question\nInput has already been classified\n7. Classification Categories\n\nThe following 15 categories form the fixed classification list:\n\nCategory\tCode\tDescription\nCODE_GENERATION\tCG\tWriting, modifying, or generating source code\nCODE_REVIEW\tCR\tReviewing, analyzing, or auditing existing code\nDEBUGGING\tDB\tIdentifying and fixing bugs, errors, or issues\nDATA_ANALYSIS\tDA\tAnalyzing, processing, or visualizing data\nFILE_OPERATIONS\tFO\tReading, writing, moving, or managing files\nDOCUMENTATION\tDC\tCreating or updating documentation\nREFACTORING\tRF\tRestructuring code without changing behavior\nTESTING\tTS\tWriting, running, or managing tests\nDEPLOYMENT\tDP\tDeploying applications or infrastructure\nRESEARCH\tRS\tInvestigating, searching, or gathering information\nCONFIGURATION\tCF\tSetting up or modifying configurations\nCOMMUNICATION\tCM\tDrafting messages, emails, or communications\nCONVERSION\tCV\tTransforming data or files between formats\nANALYSIS\tAN\tGeneral analysis not covered by other categories\nPLANNING\tPL\tCreating plans, strategies, or roadmaps\n\nCategory Priority Order (for tie-breaking):\n\nDEBUGGING (highest - immediate attention needed)\nDEPLOYMENT\nCODE_GENERATION\nREFACTORING\nTESTING\nCODE_REVIEW\nDATA_ANALYSIS\nCONFIGURATION\nFILE_OPERATIONS\nCONVERSION\nDOCUMENTATION\nRESEARCH\nANALYSIS\nCOMMUNICATION\nPLANNING (lowest)\n8. Category Boundary Definitions\nCODE_GENERATION vs CODE_REVIEW\nCODE_GENERATION: Input requests creating new code or modifying existing code\nCODE_REVIEW: Input requests analysis of existing code without modifications\nBoundary: If modification is implied, classify as CODE_GENERATION\nCODE_GENERATION vs REFACTORING\nCODE_GENERATION: Creating new functionality or features\nREFACTORING: Restructuring existing code without new functionality\nBoundary: \"Improve performance\" without feature change = REFACTORING\nDEBUGGING vs CODE_GENERATION\nDEBUGGING: Input explicitly mentions errors, bugs, or failures\nCODE_GENERATION: Input requests new features without error context\nBoundary: \"Fix this bug\" = DEBUGGING; \"Add error handling\" = CODE_GENERATION\nDATA_ANALYSIS vs ANALYSIS\nDATA_ANALYSIS: Input involves data processing, statistics, or visualization\nANALYSIS: Input involves general analysis of concepts, requirements, or situations\nBoundary: If data files/sets are mentioned = DATA_ANALYSIS\nFILE_OPERATIONS vs CONVERSION\nFILE_OPERATIONS: Input requests file management (read, write, move, delete)\nCONVERSION: Input requests format transformation between file types\nBoundary: \"Convert X to Y format\" = CONVERSION; \"Read file X\" = FILE_OPERATIONS\nRESEARCH vs ANALYSIS\nRESEARCH: Input requests information gathering or investigation\nANALYSIS: Input requests evaluation or assessment of known information\nBoundary: \"Find information about X\" = RESEARCH; \"Evaluate X\" = ANALYSIS\nTESTING vs CODE_GENERATION\nTESTING: Input specifically requests test creation or test execution\nCODE_GENERATION: Input requests production code\nBoundary: \"Write tests for X\" = TESTING; \"Write X with tests\" = CODE_GENERATION (primary)\nCONFIGURATION vs DEPLOYMENT\nCONFIGURATION: Input requests setup of settings or configurations\nDEPLOYMENT: Input requests deployment to environments or infrastructure\nBoundary: Local setup = CONFIGURATION; Remote/environment setup = DEPLOYMENT\n9. Single-Primary-Category Rule\n\nRule: Every classified input MUST have exactly one primary category.\n\nEnforcement Rules\nNo Multi-Category Output: Never output multiple primary categories\nTie-Breaking Required: When multiple categories match, apply tie-breaking logic\nCategory Exclusivity: Primary category is mutually exclusive with other primary categories\nTie-Breaking Logic\n\nWhen input matches multiple categories, apply in order:\n\nStep 1: Keyword Dominance\n\nCount explicit keyword matches for each candidate category\nCategory with highest keyword count wins\nIf tied, proceed to Step 2\n\nStep 2: Action Verb Analysis\n\nIdentify primary action verb in input\nMap action verb to category using verb-to-category mapping\nIf still tied, proceed to Step 3\n\nStep 3: Priority Order\n\nApply category priority order (see Section 7)\nHigher priority category wins\n\nStep 4: Default Fallback\n\nIf all steps fail, default to ANALYSIS category\nTie-Breaking Example\n\nInput: \"Debug and fix the error in the authentication code, then add logging\"\n\nKeywords: DEBUGGING (debug, fix, error), CODE_GENERATION (add, logging)\nAction Verb: \"Debug\" (primary action) → DEBUGGING\nResult: DEBUGGING (primary), secondary_tags: [CODE_GENERATION]\n10. Secondary Tag Rules\n\nSecondary tags provide additional context without affecting primary routing.\n\nRules\nMaximum 3 Tags: No more than 3 secondary tags per classification\nNo Primary Duplicate: Secondary tags cannot duplicate primary category\nRelated Categories Only: Secondary tags must be from the fixed category list\nRelevance Threshold: Only add tags with >50% keyword match confidence\nSecondary Tag Selection Process\nIdentify all categories with keyword matches (excluding primary)\nCalculate relevance score for each (0.0-1.0)\nSort by relevance score (descending)\nSelect top 3 with score > 0.5\nAdd to ClassificationResult\nWhen to Apply Secondary Tags\nInput contains multiple distinct sub-tasks\nInput references multiple technology domains\nInput implies follow-up work in other categories\nInput has context from previous interactions in different categories\nWhen NOT to Apply Secondary Tags\nInput is single-focused with no additional context\nRelevance scores are below threshold (0.5)\nWould duplicate primary category\nWould exceed 3-tag limit\nSecondary Tag Examples\nInput\tPrimary\tSecondary Tags\n\"Debug the API and update the docs\"\tDEBUGGING\t[DOCUMENTATION]\n\"Refactor the auth module and add tests\"\tREFACTORING\t[TESTING]\n\"Analyze sales data and create a report\"\tDATA_ANALYSIS\t[DOCUMENTATION, COMMUNICATION]\n\"Fix this bug\"\tDEBUGGING\t[]\nClassification Models Reference\n\nFor detailed complexity, risk, confidence, and state transition models, see classification-models.md.\n\nFor system integration, failure conditions, logging, and examples, see system-integration.md."
  },
  "trust": {
    "sourceLabel": "tencent",
    "provenanceUrl": "https://clawhub.ai/omprasad122007-rgb/input-classification-v1",
    "publisherUrl": "https://clawhub.ai/omprasad122007-rgb/input-classification-v1",
    "owner": "omprasad122007-rgb",
    "version": "1.0.0",
    "license": null,
    "verificationStatus": "Indexed source record"
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/input-classification-v1",
    "downloadUrl": "https://openagent3.xyz/downloads/input-classification-v1",
    "agentUrl": "https://openagent3.xyz/skills/input-classification-v1/agent",
    "manifestUrl": "https://openagent3.xyz/skills/input-classification-v1/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/input-classification-v1/agent.md"
  }
}