documcp

# How to Prompt DocuMCP Effectively This guide shows you how to interact with DocuMCP using effective prompts to get the best results from the system. ## Quick Start DocuMCP responds to natural language prompts. Here are the most common patterns: ### Basic Analysis ``` analyze my repository for documentation needs ``` ### Get Recommendations ``` what static site generator should I use for my project? ``` ### Deploy Documentation ``` set up GitHub Pages deployment for my docs ``` ## Available Tools DocuMCP provides several tools you can invoke through natural prompts: ### 1. Repository Analysis **Purpose**: Analyze your project structure, dependencies, and documentation needs. **Example Prompts**: - "Analyze my repository structure" - "What documentation gaps do I have?" - "Examine my project for documentation opportunities" **What it returns**: Project analysis with language detection, dependency mapping, and complexity assessment. ### 2. SSG Recommendations **Purpose**: Get intelligent static site generator recommendations based on your project. **Example Prompts**: - "Recommend a static site generator for my TypeScript project" - "Which SSG works best with my Python documentation?" - "Compare documentation tools for my project" **What it returns**: Weighted recommendations with justifications for Jekyll, Hugo, Docusaurus, MkDocs, or Eleventy. ### 3. Configuration Generation **Purpose**: Generate SSG-specific configuration files. **Example Prompts**: - "Generate a Hugo config for my project" - "Create MkDocs configuration files" - "Set up Docusaurus for my documentation" **What it returns**: Ready-to-use configuration files optimized for your project. ### 4. Documentation Structure **Purpose**: Create Diataxis-compliant documentation structure. **Example Prompts**: - "Set up documentation structure following Diataxis" - "Create organized docs folders for my project" - "Build a comprehensive documentation layout" **What it returns**: Organized folder structure with templates following documentation best practices. ### 5. GitHub Pages Deployment **Purpose**: Automate GitHub Pages deployment workflows. **Example Prompts**: - "Deploy my docs to GitHub Pages" - "Set up automated documentation deployment" - "Create GitHub Actions for my documentation site" **What it returns**: GitHub Actions workflows configured for your chosen SSG. ### 6. Deployment Verification **Purpose**: Verify and troubleshoot GitHub Pages deployments. **Example Prompts**: - "Check if my GitHub Pages deployment is working" - "Troubleshoot my documentation deployment" - "Verify my docs site is live" **What it returns**: Deployment status and troubleshooting recommendations. ## Advanced Prompting Techniques ### Chained Operations You can chain multiple operations in a single conversation: ``` 1. First analyze my repository 2. Then recommend the best SSG 3. Finally set up the deployment workflow ``` ### Specific Requirements Be specific about your needs: ``` I need a documentation site that: - Works with TypeScript - Supports API documentation - Has good search functionality - Deploys automatically on commits ``` ### Context-Aware Requests Reference previous analysis: ``` Based on the analysis you just did, create the documentation structure and deploy it to GitHub Pages ``` ## Best Practices ### 1. Start with Analysis Always begin with repository analysis to get tailored recommendations: ``` analyze my project for documentation needs ``` ### 2. Be Specific About Goals Tell DocuMCP what you're trying to achieve: - "I need developer documentation for my API" - "I want user guides for my application" - "I need project documentation for contributors" ### 3. Specify Constraints Mention any limitations or preferences: - "I prefer minimal setup" - "I need something that works with our CI/CD pipeline" - "I want to use our existing design system" ### 4. Ask for Explanations Request reasoning behind recommendations: ``` why did you recommend Hugo over Jekyll for my project? ``` ### 5. Iterate and Refine Use follow-up prompts to refine results: ``` can you modify the GitHub Actions workflow to also run tests? ``` ## Common Workflows ### Complete Documentation Setup ``` 1. "Analyze my repository for documentation needs" 2. "Recommend the best static site generator for my project" 3. "Generate configuration files for the recommended SSG" 4. "Set up Diataxis-compliant documentation structure" 5. "Deploy everything to GitHub Pages" ``` ### Documentation Audit ``` 1. "Analyze my existing documentation" 2. "What gaps do you see in my current docs?" 3. "How can I improve my documentation structure?" ``` ### Deployment Troubleshooting ``` 1. "My GitHub Pages site isn't working" 2. "Check my deployment configuration" 3. "Help me fix the build errors" ``` ## Memory and Context DocuMCP remembers context within a conversation, so you can: - Reference previous analysis results - Build on earlier recommendations - Chain operations together seamlessly Example conversation flow: ``` User: "analyze my repository" DocuMCP: [provides analysis] User: "based on that analysis, what SSG do you recommend?" DocuMCP: [provides recommendation using analysis context] User: "set it up with that recommendation" DocuMCP: [configures the recommended SSG] ``` ## Troubleshooting Prompts If you're not getting the results you expect, try: ### More Specific Prompts Instead of: "help with docs" Try: "analyze my TypeScript project and recommend documentation tools" ### Context Setting Instead of: "set up deployment" Try: "set up GitHub Pages deployment for the MkDocs site we just configured" ### Direct Tool Requests If you know exactly what you want: - "use the analyze_repository tool on my current directory" - "run the recommend_ssg tool with my project data" ## Getting Help If you need assistance with prompting: - Ask DocuMCP to explain available tools: "what can you help me with?" - Request examples: "show me example prompts for documentation setup" - Ask for clarification: "I don't understand the recommendation, can you explain?" Remember: DocuMCP is designed to understand natural language, so don't hesitate to ask questions in your own words!