UNPKG

context-forge

Version:

AI orchestration platform with autonomous teams, enhancement planning, migration tools, 25+ slash commands, checkpoints & hooks. Multi-IDE: Claude, Cursor, Windsurf, Cline, Copilot

291 lines (225 loc) 6.54 kB
# Retrofit Guide - Context Forge ## What is Retrofit? The retrofit feature (via the `analyze` command) allows you to add AI-optimized documentation and configuration to existing projects. It's designed to seamlessly integrate Context Forge into established codebases without disrupting your current workflow. ## When to Use Retrofit Use the `analyze` command when you have: - An existing project that needs AI documentation - A legacy codebase you want to modernize with AI assistance - A project you're inheriting and need to understand - A codebase that lacks proper documentation ## Step-by-Step Workflow ### 1. Initial Analysis ```bash # Basic analysis context-forge analyze # With specific output directory context-forge analyze -o ./ai-docs # Target specific IDEs context-forge analyze -i claude,cursor ``` ### 2. What Gets Analyzed The analyzer examines: - **Project Structure**: Directory layout, file organization - **Tech Stack**: Frameworks, libraries, dependencies - **Code Patterns**: Architecture, design patterns - **Quality Metrics**: Complexity, test coverage - **Documentation**: Existing docs, comments, READMEs ### 3. Interactive Configuration During analysis, you'll be asked about: 1. **Project Confirmation** - Verify detected tech stack - Confirm project type - Set project name and description 2. **Documentation Preferences** - Choose documentation depth - Select target AI IDEs - Enable advanced features 3. **Integration Options** - Slash commands (Claude Code) - Hooks and automation - Checkpoint system ### 4. Generated Outputs #### Standard Files ``` your-project/ ├── CLAUDE.md # Main context file ├── PRPs/ # Implementation guides │ └── retrofit/ # Retrofit-specific PRPs └── .gitignore # Updated with AI files ``` #### Claude Code Specific ``` .claude/ ├── commands/ # Custom slash commands │ ├── run-tests.md │ ├── check-lint.md │ └── project-status.md ├── hooks/ # Automation scripts │ ├── pre-commit.py │ └── post-test.py └── docs/ # Extended documentation ``` ## Common Retrofit Patterns ### 1. Minimal Integration For projects that just need basic AI context: ```bash context-forge analyze --quick ``` ### 2. Full Integration For comprehensive AI assistance: ```bash context-forge analyze --deep ``` ### 3. Incremental Adoption Start small and expand: ```bash # Phase 1: Basic documentation context-forge analyze --quick # Phase 2: Add PRPs context-forge analyze --prp # Phase 3: Enable automation context-forge analyze --hooks --checkpoints ``` ## Handling Special Cases ### 1. Monorepos ```bash # Analyze from monorepo root cd my-monorepo context-forge analyze # Or analyze specific package cd packages/api context-forge analyze --scope package ``` ### 2. Legacy Codebases ```bash # Use legacy mode for older projects context-forge analyze --legacy # This enables: # - Broader framework detection # - Legacy pattern recognition # - Compatibility documentation ``` ### 3. Mixed Tech Stacks ```bash # Explicitly specify primary stack context-forge analyze --primary-stack react # For full-stack projects context-forge analyze --full-stack ``` ## Advanced Usage ### 1. Custom Analysis Depth ```bash # Shallow analysis (faster) context-forge analyze --depth shallow # Deep analysis (comprehensive) context-forge analyze --depth deep # With AI-powered insights context-forge analyze --ai-insights ``` ### 2. Excluding Paths ```bash # Exclude specific directories context-forge analyze --exclude node_modules,dist,build # Use .cfignore file echo "*.test.js" >> .cfignore context-forge analyze ``` ### 3. Framework-Specific Options ```bash # React projects context-forge analyze --framework react --detect-patterns hooks,context # Node.js APIs context-forge analyze --framework express --detect-patterns middleware,routes ``` ## Integration with Existing Tools ### 1. Preserving Existing Documentation Context Forge appends to existing files rather than overwriting: - Existing CLAUDE.md files are backed up - READMEs are preserved - Comments in code remain untouched ### 2. Git Integration ```bash # See what will be added context-forge analyze --dry-run # Commit AI documentation separately git add .claude CLAUDE.md PRPs/ git commit -m "Add AI development context" ``` ### 3. CI/CD Compatibility Add to your CI pipeline: ```yaml # .github/workflows/ai-docs.yml - name: Update AI Documentation run: | npx context-forge analyze --ci git add . git commit -m "Update AI context" || true ``` ## Best Practices ### 1. Initial Setup - Run analysis on a clean working directory - Review detected tech stack carefully - Enable all Claude Code features if using Claude ### 2. Maintenance - Re-run analysis after major changes - Update PRPs when adding features - Keep slash commands synchronized ### 3. Team Adoption - Share generated docs with team - Document custom patterns in ai_docs/ - Create team-specific PRPs ## Example: Retrofitting a React App ```bash # 1. Navigate to your React project cd my-react-app # 2. Run comprehensive analysis context-forge analyze -i claude --deep # 3. During prompts: # - Confirm React + TypeScript stack # - Enable hooks, commands, checkpoints # - Select comprehensive documentation # 4. Review generated files cat CLAUDE.md ls -la .claude/commands/ ls -la PRPs/ # 5. Test with Claude Code # Open project in Claude Code # Try: /project-status # Try: /run-tests ``` ## Troubleshooting ### Detection Issues ```bash # Force framework detection context-forge analyze --force-detect # Specify frameworks manually context-forge analyze --frameworks react,typescript,jest ``` ### Performance ```bash # For large codebases context-forge analyze --parallel # Limit file scanning context-forge analyze --max-files 1000 ``` ### Integration Problems ```bash # Check compatibility context-forge analyze --check-only # Generate report context-forge analyze --report-only ``` ## Next Steps After retrofitting your project: 1. **Test AI Integration** - Open in your AI IDE - Test slash commands - Verify context understanding 2. **Enhance Documentation** - Add project-specific PRPs - Document unique patterns - Create custom commands 3. **Plan Improvements** - Use `enhance` for new features - Consider `migrate` for tech updates - Keep context current with regular analysis