Project Memory MCP

import { TASK_JSON_SCHEMA } from '../schemas/task-schema.js'; /** * Organize/migrate existing CLAUDE.md into project-memory structure * This is a one-time migration tool for existing projects */ export const ORGANIZE_PROMPT = ` # Organize CLAUDE.md into Project Memory You are helping migrate an existing project's CLAUDE.md into the project-memory system. ## Goal Extract architecture, conventions, commands, tasks, and specs from CLAUDE.md and organize them into \`.project-memory/\` while keeping minimal references in CLAUDE.md. --- ## Step 1: Initialize Project Memory (if not exists) Check if \`.project-memory/\` folder exists: \`\`\`bash ls -la .project-memory 2>/dev/null || echo "Not initialized" \`\`\` If not initialized, run the \`project-memory init\` tool first, then return to this organize workflow. --- ## Step 2: Read and Analyze CLAUDE.md Use \`Read\` tool to read the current CLAUDE.md (or claude.md, check both): **Identify these sections:** 1. **Architecture/Design** - System design, project structure, technical decisions 2. **Conventions/Standards** - Coding standards, style guides, patterns 3. **Commands** - Common commands, scripts, build/test/deploy instructions 4. **Tasks/TODOs** - Active tasks, task lists, work items 5. **Specs/Plans** - Implementation plans, feature specs, requirements **Mark line numbers** for each section to extract. --- ## Step 3: Create Migration Plan Present a clear migration plan to user using \`AskUserQuestion\`: \`\`\` Found the following sections in CLAUDE.md: 📐 Architecture (lines 45-120, 75 lines) → Will move to: .project-memory/architecture.md 📋 Coding Standards (lines 121-180, 60 lines) → Will move to: .project-memory/conventions.md ⚡ Common Commands (lines 181-210, 30 lines) → Will move to: .project-memory/useful-commands.md ✅ Active Tasks (lines 15-44, task list format) → Will convert to: .project-memory/tasks/tasks-active.json 📄 Feature Spec: Authentication (lines 220-280) → Will move to: .project-memory/specs/authentication.md After migration, CLAUDE.md will have minimal references pointing to these files. Proceed with migration? \`\`\` **IMPORTANT:** Get explicit user approval before proceeding. --- ## Step 4: Migrate Content After approval, migrate each section: ### 4.1: Migrate Architecture If architecture section found: 1. Read the architecture content from CLAUDE.md 2. Check if \`.project-memory/architecture.md\` exists 3. If exists, use \`Edit\` to append/merge; if not, use \`Write\` to create 4. Format properly with markdown headers ### 4.2: Migrate Conventions If conventions section found: 1. Read the conventions content from CLAUDE.md 2. Check if \`.project-memory/conventions.md\` exists 3. If exists, merge with existing; if not, create new 4. Organize into logical sections (File Naming, Code Style, Testing, etc.) ### 4.3: Migrate Commands If commands section found: 1. Read the commands content from CLAUDE.md 2. Check if \`.project-memory/useful-commands.md\` exists 3. If exists, merge; if not, create 4. Organize by category (Development, Build, Test, Deploy, etc.) ### 4.4: Migrate Tasks If tasks/todos found: 1. Read the task content from CLAUDE.md 2. Parse into structured task format: ${TASK_JSON_SCHEMA} 3. Read existing \`.project-memory/tasks/tasks-active.json\` 4. Merge new tasks with existing (avoid duplicate IDs) 5. Update tasks-active.json with \`Write\` or \`Edit\` **Task Parsing Rules:** - Generate unique IDs (TASK-001, TASK-002, etc.) - Set status to "pending" unless marked done - Extract acceptance criteria if present - Identify dependencies between tasks ### 4.5: Migrate Specs If spec/plan sections found: 1. For each spec section, create a separate .md file in \`.project-memory/specs/\` 2. Use descriptive filenames: \`feature-auth.md\`, \`api-redesign.md\`, etc. 3. Preserve full spec content 4. Mark as immutable (add note at top: "This is an immutable spec") --- ## Step 5: Update CLAUDE.md with References After migrating content, update CLAUDE.md to replace verbose sections with minimal references: Use \`Edit\` tool to replace each migrated section with a reference: **Architecture section** → Replace with: \`\`\`markdown ## Architecture See \`.project-memory/architecture.md\` for system design and project structure. \`\`\` **Conventions section** → Replace with: \`\`\`markdown ## Coding Conventions See \`.project-memory/conventions.md\` for coding standards and patterns. \`\`\` **Commands section** → Replace with: \`\`\`markdown ## Common Commands See \`.project-memory/useful-commands.md\` for build, test, and deployment commands. \`\`\` **Tasks section** → Replace with: \`\`\`markdown ## Current Tasks See \`.project-memory/tasks/tasks-active.json\` for active work items and priorities. \`\`\` **Specs section** → Replace with: \`\`\`markdown ## Specifications See \`.project-memory/specs/\` for detailed feature specifications and implementation plans. \`\`\` **CRITICAL:** Keep the "IMPORTANT: Project Memory System" section that was added during init! --- ## Step 6: Verify and Summarize 1. **Verify all files created:** - List \`.project-memory/\` contents using \`Bash\` - Confirm all migrations successful 2. **Show summary** to user: \`\`\` ✅ Migration complete! Migrated content: - Architecture (75 lines) → .project-memory/architecture.md - Conventions (60 lines) → .project-memory/conventions.md - Commands (30 lines) → .project-memory/useful-commands.md - Tasks (5 items) → .project-memory/tasks/tasks-active.json - Specs (1 file) → .project-memory/specs/authentication.md CLAUDE.md reduced from 350 lines to 120 lines. Next steps: - Review migrated files in .project-memory/ - Use 'project-memory parse-tasks' for any new specs - Use 'project-memory review' before commits - Use 'project-memory sync' after commits \`\`\` --- ## Important Notes 1. **Always get user approval** before making any changes 2. **Preserve all content** - don't delete anything, only move/reorganize 3. **Keep references in CLAUDE.md** - never leave it disconnected from project-memory 4. **Merge intelligently** - if project-memory files already exist, merge rather than overwrite 5. **Maintain markdown formatting** - ensure all files are properly formatted 6. **Generate valid JSON** for tasks file You're done! The project is now organized with project-memory structure. `.trim();