{
  "schemaVersion": "1.0",
  "item": {
    "slug": "agent-docs",
    "name": "Agent Docs",
    "source": "tencent",
    "type": "skill",
    "category": "效率提升",
    "sourceUrl": "https://clawhub.ai/tylervovan/agent-docs",
    "canonicalUrl": "https://clawhub.ai/tylervovan/agent-docs",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadMode": "redirect",
    "downloadUrl": "/downloads/agent-docs",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=agent-docs",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "installMethod": "Manual import",
    "extraction": "Extract archive",
    "prerequisites": [
      "OpenClaw"
    ],
    "packageFormat": "ZIP package",
    "includedAssets": [
      "SKILL.md",
      "references/advanced-patterns.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",
      "status": "healthy",
      "reason": "direct_download_ok",
      "recommendedAction": "download",
      "checkedAt": "2026-04-23T16:43:11.935Z",
      "expiresAt": "2026-04-30T16:43:11.935Z",
      "httpStatus": 200,
      "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=4claw-imageboard",
      "contentType": "application/zip",
      "probeMethod": "head",
      "details": {
        "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=4claw-imageboard",
        "contentDisposition": "attachment; filename=\"4claw-imageboard-1.0.1.zip\"",
        "redirectLocation": null,
        "bodySnippet": null
      },
      "scope": "source",
      "summary": "Source download looks usable.",
      "detail": "Yavira can redirect you to the upstream package for this source.",
      "primaryActionLabel": "Download for OpenClaw",
      "primaryActionHref": "/downloads/agent-docs"
    },
    "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/agent-docs",
    "agentPageUrl": "https://openagent3.xyz/skills/agent-docs/agent",
    "manifestUrl": "https://openagent3.xyz/skills/agent-docs/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/agent-docs/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": "Agent Docs",
        "body": "Write documentation that AI agents can efficiently consume. Based on Vercel benchmarks and industry standards (AGENTS.md, llms.txt, CLAUDE.md)."
      },
      {
        "title": "The Hybrid Context Hierarchy",
        "body": "Three-layer architecture for optimal agent performance:"
      },
      {
        "title": "Layer 1: Constitution (Inline)",
        "body": "Always in context. 2,000–4,000 tokens max.\n\n# AGENTS.md\n> Context: Next.js 16 | Tailwind | Supabase\n\n## 🚨 CRITICAL\n- NO SECRETS in output\n- Use `app/` directory ONLY\n\n## 📚 DOCS INDEX (use read_file)\n- Auth: `docs/auth/llms.txt`\n- DB: `docs/db/schema.md`\n\nInclude:\n\nSecurity rules, architecture constraints\nBuild/test/lint commands (top for primacy bias)\nDocumentation map (where to find more)"
      },
      {
        "title": "Layer 2: Reference Library (Local Retrieval)",
        "body": "Fetched on demand. 1K–5K token chunks.\n\nFramework-specific guides\nDetailed style guides\nAPI schemas"
      },
      {
        "title": "Layer 3: Research Assistant (External)",
        "body": "Gated by allow-lists. Edge cases only.\n\nLatest library updates\nStack Overflow for obscure errors\nThird-party llms.txt"
      },
      {
        "title": "Why This Works",
        "body": "Vercel Benchmark (2026):\n\nApproachPass RateTool-based retrieval53%Retrieval + prompting79%Inline AGENTS.md100%\n\nRoot cause: Meta-cognitive failure. Agents don't know what they don't know—they assume training data is sufficient. Inline docs bypass this entirely."
      },
      {
        "title": "1. Compressed Index > Full Docs",
        "body": "An 8KB compressed index outperforms a 40KB full dump.\n\nCompress to:\n\nFile paths (where code lives)\nFunction signatures (names + types only)\nNegative constraints (\"Do NOT use X\")"
      },
      {
        "title": "2. Structure for Chunking",
        "body": "RAG systems split at headers. Each section must be self-contained:\n\n## Database Setup          ← Chunk boundary\n\nPrerequisites: PostgreSQL 14+\n\n1. Create database...\n\nRules:\n\nFront-load key info (chunkers truncate)\nDescriptive headers (agents search by header text)"
      },
      {
        "title": "3. Inline Over Links",
        "body": "Agents can't autonomously browse. Each link = tool call + latency + potential failure.\n\nApproachToken LoadAgent SuccessFull inline~12K✅ HighLinks only~2K❌ Requires fetchingHybrid~4K base✅ Best of both"
      },
      {
        "title": "4. The \"Lost in the Middle\" Problem",
        "body": "LLMs have U-shaped attention:\n\nStrong: Start of context (primacy)\nStrong: End of context (recency)\nWeak: Middle of context\n\nSolution: Put critical rules at TOP of AGENTS.md. Governance first, details later."
      },
      {
        "title": "5. Signal-to-Noise Ratio",
        "body": "Strip everything that isn't essential:\n\nNo \"Welcome to...\" preambles\nNo marketing text\nNo changelogs in core docs\n\nFormats like llms.txt and AGENTS.md mechanically increase SNR."
      },
      {
        "title": "llms.txt Standard",
        "body": "Machine-readable doc index for agents:\n\n# Project Name\n\n> One-line project description.\n\n## Authentication\n\n- [Setup](docs/auth/setup.md): Environment vars and init\n- [Server](docs/auth/server.md): Cookie handling\n\n## Database\n\n- [Schema](docs/db/schema.md): Full Prisma schema\n\nLocation: /llms.txt at domain root\nCompanion: /llms-full.txt — full concatenated docs, HTML stripped"
      },
      {
        "title": "Inline = Trusted",
        "body": "AGENTS.md is part of your codebase. Controlled, version-pinned."
      },
      {
        "title": "External = Attack Surface",
        "body": "Indirect prompt injection via hidden text\nSSRF risks if agents can browse freely\nDependency on external uptime\n\nMitigation: Domain allow-lists, human-in-the-loop for external retrieval."
      },
      {
        "title": "Anti-Patterns",
        "body": "Pasting 50 pages — triggers \"Lost in the Middle\"\n\"See external docs\" — agents can't browse autonomously\nGeneric advice — \"Write clean code\" (use specific constraints)\nTOC-only docs — indexes without content\nTrusting retrieval alone — 53% vs 100% pass rate"
      },
      {
        "title": "Advanced Patterns",
        "body": "For detailed guidance on RAG optimization, multi-framework docs, and API templates, see references/advanced-patterns.md."
      },
      {
        "title": "Validation Checklist",
        "body": "Critical governance at TOP of doc\n Total inline context under 4K tokens\n Each H2 section self-contained\n No external links without inline summary\n Negative constraints explicit (\"Do NOT...\")\n File paths and signatures, not full code"
      }
    ],
    "body": "Agent Docs\n\nWrite documentation that AI agents can efficiently consume. Based on Vercel benchmarks and industry standards (AGENTS.md, llms.txt, CLAUDE.md).\n\nThe Hybrid Context Hierarchy\n\nThree-layer architecture for optimal agent performance:\n\nLayer 1: Constitution (Inline)\n\nAlways in context. 2,000–4,000 tokens max.\n\n# AGENTS.md\n> Context: Next.js 16 | Tailwind | Supabase\n\n## 🚨 CRITICAL\n- NO SECRETS in output\n- Use `app/` directory ONLY\n\n## 📚 DOCS INDEX (use read_file)\n- Auth: `docs/auth/llms.txt`\n- DB: `docs/db/schema.md`\n\n\nInclude:\n\nSecurity rules, architecture constraints\nBuild/test/lint commands (top for primacy bias)\nDocumentation map (where to find more)\nLayer 2: Reference Library (Local Retrieval)\n\nFetched on demand. 1K–5K token chunks.\n\nFramework-specific guides\nDetailed style guides\nAPI schemas\nLayer 3: Research Assistant (External)\n\nGated by allow-lists. Edge cases only.\n\nLatest library updates\nStack Overflow for obscure errors\nThird-party llms.txt\nWhy This Works\n\nVercel Benchmark (2026):\n\nApproach\tPass Rate\nTool-based retrieval\t53%\nRetrieval + prompting\t79%\nInline AGENTS.md\t100%\n\nRoot cause: Meta-cognitive failure. Agents don't know what they don't know—they assume training data is sufficient. Inline docs bypass this entirely.\n\nCore Principles\n1. Compressed Index > Full Docs\n\nAn 8KB compressed index outperforms a 40KB full dump.\n\nCompress to:\n\nFile paths (where code lives)\nFunction signatures (names + types only)\nNegative constraints (\"Do NOT use X\")\n2. Structure for Chunking\n\nRAG systems split at headers. Each section must be self-contained:\n\n## Database Setup          ← Chunk boundary\n\nPrerequisites: PostgreSQL 14+\n\n1. Create database...\n\n\nRules:\n\nFront-load key info (chunkers truncate)\nDescriptive headers (agents search by header text)\n3. Inline Over Links\n\nAgents can't autonomously browse. Each link = tool call + latency + potential failure.\n\nApproach\tToken Load\tAgent Success\nFull inline\t~12K\t✅ High\nLinks only\t~2K\t❌ Requires fetching\nHybrid\t~4K base\t✅ Best of both\n4. The \"Lost in the Middle\" Problem\n\nLLMs have U-shaped attention:\n\nStrong: Start of context (primacy)\nStrong: End of context (recency)\nWeak: Middle of context\n\nSolution: Put critical rules at TOP of AGENTS.md. Governance first, details later.\n\n5. Signal-to-Noise Ratio\n\nStrip everything that isn't essential:\n\nNo \"Welcome to...\" preambles\nNo marketing text\nNo changelogs in core docs\n\nFormats like llms.txt and AGENTS.md mechanically increase SNR.\n\nllms.txt Standard\n\nMachine-readable doc index for agents:\n\n# Project Name\n\n> One-line project description.\n\n## Authentication\n\n- [Setup](docs/auth/setup.md): Environment vars and init\n- [Server](docs/auth/server.md): Cookie handling\n\n## Database\n\n- [Schema](docs/db/schema.md): Full Prisma schema\n\n\nLocation: /llms.txt at domain root Companion: /llms-full.txt — full concatenated docs, HTML stripped\n\nSecurity Considerations\nInline = Trusted\n\nAGENTS.md is part of your codebase. Controlled, version-pinned.\n\nExternal = Attack Surface\nIndirect prompt injection via hidden text\nSSRF risks if agents can browse freely\nDependency on external uptime\n\nMitigation: Domain allow-lists, human-in-the-loop for external retrieval.\n\nAnti-Patterns\nPasting 50 pages — triggers \"Lost in the Middle\"\n\"See external docs\" — agents can't browse autonomously\nGeneric advice — \"Write clean code\" (use specific constraints)\nTOC-only docs — indexes without content\nTrusting retrieval alone — 53% vs 100% pass rate\nAdvanced Patterns\n\nFor detailed guidance on RAG optimization, multi-framework docs, and API templates, see references/advanced-patterns.md.\n\nValidation Checklist\n Critical governance at TOP of doc\n Total inline context under 4K tokens\n Each H2 section self-contained\n No external links without inline summary\n Negative constraints explicit (\"Do NOT...\")\n File paths and signatures, not full code"
  },
  "trust": {
    "sourceLabel": "tencent",
    "provenanceUrl": "https://clawhub.ai/tylervovan/agent-docs",
    "publisherUrl": "https://clawhub.ai/tylervovan/agent-docs",
    "owner": "tylervovan",
    "version": "1.0.0",
    "license": null,
    "verificationStatus": "Indexed source record"
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/agent-docs",
    "downloadUrl": "https://openagent3.xyz/downloads/agent-docs",
    "agentUrl": "https://openagent3.xyz/skills/agent-docs/agent",
    "manifestUrl": "https://openagent3.xyz/skills/agent-docs/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/agent-docs/agent.md"
  }
}