# Send Documentation-Accurate Code Generation to your agent Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually. ## Fast path - 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. ## Suggested prompts ### New install ```text 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. ``` ### Upgrade existing ```text 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. ``` ## Machine-readable fields ```json { "schemaVersion": "1.0", "item": { "slug": "doc-accurate-codegen", "name": "Documentation-Accurate Code Generation", "source": "tencent", "type": "skill", "category": "效率提升", "sourceUrl": "https://clawhub.ai/tobisamaa/doc-accurate-codegen", "canonicalUrl": "https://clawhub.ai/tobisamaa/doc-accurate-codegen", "targetPlatform": "OpenClaw" }, "install": { "downloadUrl": "/downloads/doc-accurate-codegen", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=doc-accurate-codegen", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "packageFormat": "ZIP package", "primaryDoc": "SKILL.md", "includedAssets": [ "SKILL.md" ], "downloadMode": "redirect", "sourceHealth": { "source": "tencent", "slug": "doc-accurate-codegen", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-04-29T18:47:32.704Z", "expiresAt": "2026-05-06T18:47:32.704Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=doc-accurate-codegen", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=doc-accurate-codegen", "contentDisposition": "attachment; filename=\"doc-accurate-codegen-1.0.0.zip\"", "redirectLocation": null, "bodySnippet": null, "slug": "doc-accurate-codegen" }, "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/doc-accurate-codegen" }, "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." ] } }, "links": { "detailUrl": "https://openagent3.xyz/skills/doc-accurate-codegen", "downloadUrl": "https://openagent3.xyz/downloads/doc-accurate-codegen", "agentUrl": "https://openagent3.xyz/skills/doc-accurate-codegen/agent", "manifestUrl": "https://openagent3.xyz/skills/doc-accurate-codegen/agent.json", "briefUrl": "https://openagent3.xyz/skills/doc-accurate-codegen/agent.md" } } ``` ## Documentation ### Documentation-Accurate Code Generation CRITICAL: This skill prevents LLM hallucination by enforcing documentation reference. ### When to Use ALWAYS when generating code ALWAYS when using APIs ALWAYS when creating configurations ALWAYS when implementing features ### Core Philosophy NEVER generate code from memory. ALWAYS reference documentation. ### The Problem LLMs hallucinate APIs that don't exist Methods get renamed or removed Parameters change or get deprecated Return types shift unexpectedly Configuration formats evolve ### The Solution Load documentation FIRST — Before writing any code Extract API signatures — Get actual method signatures Generate from docs — Use real API data Validate against docs — Check generated code matches Reference tracking — Document which docs were used ### Workflow 1. IDENTIFY → What code/API/tool is needed? 2. LOCATE → Find documentation source 3. LOAD → Fetch and parse documentation 4. EXTRACT → Pull API signatures, parameters, examples 5. GENERATE → Create code using actual docs 6. VALIDATE → Check code matches documentation 7. REFERENCE → Track what docs were used ### 1. OpenClaw Internal Docs Location: C:\\Users\\clipp\\AppData\\Roaming\\npm\\node_modules\\openclaw\\docs Access: read tool Use: For OpenClaw-specific APIs, tools, skills ### 2. Tool Documentation Tool help: --help flags Man pages: man Official docs: Use web_fetch to get docs ### 3. API Documentation Official docs: Use web_fetch OpenAPI specs: Parse and reference Package docs: npm, pip, cargo docs ### 4. Code Examples Existing code: Read similar implementations Tests: Check test files for usage patterns Examples: Find working code samples ### Step 1: Documentation Discovery # For OpenClaw tools read("openclaw-docs-path/tool-name.md") # For external tools web_fetch("https://docs.tool.com/api") # For local tools exec("tool --help") ### Step 2: API Signature Extraction # Extract: - Method names - Parameters (names, types, required/optional) - Return types - Error handling - Examples - Version information ### Step 3: Code Generation # Generate code using actual API data def generate_from_docs(api_docs): # Use real method names # Use real parameter names # Use real return types # Include error handling from docs # Add docstrings from docs pass ### Step 4: Validation def validate_against_docs(code, api_docs): # Check method names match # Check parameter names match # Check types match # Check return types match # Verify no hallucinated methods pass ### Quick Actions codegen — Generate code with doc reference validate — Check code against docs doc-lookup — Load and display documentation api-extract — Extract API signatures ### Usage Examples "Generate code to use the OpenClaw sessions_spawn tool" # Process: Load docs → Extract API → Generate → Validate "Create a Python script using the requests library" # Process: Fetch requests docs → Extract API → Generate → Validate "Write configuration for OpenClaw channels" # Process: Load config docs → Extract format → Generate → Validate ### 1. Method Name Validation Check method exists in docs Verify spelling matches exactly Confirm method is not deprecated ### 2. Parameter Validation All required parameters present Parameter names match docs exactly Parameter types match docs Optional parameters marked correctly ### 3. Return Type Validation Return type matches docs Error types match docs Edge cases handled ### 4. Configuration Validation Keys match documentation Value types match schema Required fields present Format matches specification ### Common Hallucination Patterns Non-existent methods — Methods that don't exist Wrong parameter names — Hallucinated parameter names Wrong types — Incorrect parameter/return types Missing error handling — Ignoring documented errors Wrong configuration format — Incorrect config structure ### Prevention Strategies Always load docs first — Never generate from memory Extract actual signatures — Don't guess API shape Validate everything — Check against real docs Reference tracking — Know which docs were used Test with real APIs — Verify code actually works ### With Other Skills Coding skill: Use this for doc-accurate code Self-evolution: Update skills with doc validation Content generation: Generate accurate code examples Research: Research APIs from actual docs ### With OpenClaw Tools read: Load internal documentation web_fetch: Fetch external documentation exec: Run tools with --help for docs edit/write: Create validated code ### Format # Code Generation Reference ## Generated Code - File: path/to/file.py - Generated: 2026-02-23 - Tool: doc-accurate-codegen ## Documentation Sources 1. OpenClaw Tool Docs: /docs/tools/exec.md 2. API Reference: https://docs.example.com/api 3. Examples: /examples/exec-usage.py ## Validation - ✅ Method names validated - ✅ Parameters validated - ✅ Return types validated - ✅ Error handling validated ## Notes - Using exec tool with sandbox mode - All parameters from official docs - Error handling from API reference ### Output Template When generating code, always include: # Code generated with documentation reference # Source: [documentation URL or path] # Validated: [timestamp] # API Version: [version if available] def function_name(): """ [Docstring from actual documentation] Source: [link to docs] Parameters: [from docs] Returns: [from docs] """ # Implementation using actual API pass ### Best Practices Docs First, Always — Never generate without loading docs Exact Matches — Use exact names, types, formats from docs Validate Everything — Check all generated code Track References — Document which docs were used Test Real APIs — Actually run the code to verify Update Regularly — Re-check docs as APIs evolve Error Handling — Include all documented errors Examples — Reference working examples from docs ### Common Pitfalls Assuming API stability — APIs change, always re-check docs Memory over docs — Trust docs, not memory Partial loading — Load complete documentation No validation — Always validate generated code Missing references — Always track doc sources ### Success Metrics Hallucination rate: 0% (all code references actual docs) Validation rate: 100% (all code validated) Reference tracking: 100% (all code has doc sources) Error rate: 0% (no API misuse) Test pass rate: 100% (all generated code works) ### 1. Automatic Doc Loading Detect what APIs are needed Automatically fetch relevant docs Cache for future use ### 2. API Change Detection Monitor docs for changes Alert when APIs change Suggest code updates ### 3. Multi-Source Validation Cross-reference multiple doc sources Detect conflicts between sources Use most authoritative source ### 4. Example Extraction Extract working examples from docs Adapt examples to specific needs Test examples before using ### Tool Documentation # Get tool help exec("tool --help") # Read tool docs read("openclaw/docs/tools/tool-name.md") # Check tool examples read("openclaw/examples/tool-usage.md") ### Skill Documentation # Read skill docs read("skills/skill-name/SKILL.md") # Check skill examples read("skills/skill-name/examples/") ### Configuration Documentation # Read config docs read("openclaw/docs/configuration.md") # Check config examples read("openclaw/examples/config/") Remember: This skill exists because LLMs hallucinate. ALWAYS use it for code generation. The only way to prevent bugs is to reference actual documentation. ## Trust - Source: tencent - Verification: Indexed source record - Publisher: tobisamaa - Version: 1.0.0 ## Source health - Status: healthy - Item download looks usable. - Yavira can redirect you to the upstream package for this item. - Health scope: item - Reason: direct_download_ok - Checked at: 2026-04-29T18:47:32.704Z - Expires at: 2026-05-06T18:47:32.704Z - Recommended action: Download for OpenClaw ## Links - [Detail page](https://openagent3.xyz/skills/doc-accurate-codegen) - [Send to Agent page](https://openagent3.xyz/skills/doc-accurate-codegen/agent) - [JSON manifest](https://openagent3.xyz/skills/doc-accurate-codegen/agent.json) - [Markdown brief](https://openagent3.xyz/skills/doc-accurate-codegen/agent.md) - [Download page](https://openagent3.xyz/downloads/doc-accurate-codegen)