Project Memory MCP

import { TASK_JSON_SCHEMA } from '../schemas/task-schema.js'; /** * Fallback prompt for parse-tasks tool * Used when project-specific parse-tasks.md doesn't exist */ export const PARSE_TASKS_PROMPT = `# Task Parsing You are helping parse tasks from a spec or implementation plan. ## Task Schema ${TASK_JSON_SCHEMA} ## Instructions ### Step 1: Detect Task Storage Structure Check .project-memory/tasks/ to determine file organization: **Single-file structure (small/medium projects):** - \`tasks-active.json\` - Contains all active tasks - \`tasks-completed.json\` - Contains all completed tasks **Multi-file structure (large projects):** - \`tasks-index.json\` - Domain registry and metadata - \`tasks-active_{domain}.json\` - Domain-specific active tasks - \`tasks-completed_{domain}.json\` - Domain-specific completed tasks If \`tasks-index.json\` exists, use multi-file structure. Otherwise use single-file. If task structure is not found, assess project size/spec complexity to choose structure. ### Step 2: Parse Tasks from Spec 1. Read the spec file from .project-memory/specs/ or from the user's message 2. Extract tasks with unique IDs (TASK-001, TASK-002, etc.) 3. Assign to domains if multi-file structure (infer from task description/type) 4. Include: title, description, acceptance criteria, dependencies, priority, subtasks if needed 5. Set specReference field to the spec file path **CRITICAL: If user references a spec file, MUST include specReference in every task** - specReference links tasks back to their source specification - Format: \`"specReference": "specs/feature-name.md"\` - This enables traceability between specs and implementation - NEVER create tasks from a spec without setting specReference ### Step 3: Check Existing Tasks & Codebase **CRITICAL: Avoid duplicates and already-implemented tasks** **Check existing tasks (both active AND completed):** For single-file structure: - Read tasks-active.json AND tasks-completed.json - Compare new tasks against both files For multi-file structure: - Read ALL \`tasks-active_{domain}.json\` AND \`tasks-completed_{domain}.json\` - Compare new tasks against all files **Check codebase for already-implemented features:** - For each parsed task, verify if it's already implemented in code - Read relevant source files to confirm implementation status - Check: Does the feature exist? Does it work as described? **OUTPUT REQUIRED - Show validation results:** \`\`\` 📋 Task Validation: ✅ NEW: TASK-001 - [title] (not in tasks, not in code) ⚠️ DUPLICATE: TASK-002 - [title] (already in tasks-active.json) ⚠️ IMPLEMENTED: TASK-003 - [title] (already in code at src/...) ❌ FALSE COMPLETE: TASK-004 - [title] (in completed tasks but NOT in code!) \`\`\` **Actions based on validation:** - **NEW** → Add to active tasks - **DUPLICATE** → Skip (already tracked) - **IMPLEMENTED** → Add directly to completed tasks (not active) - **FALSE COMPLETE** → Flag to user: "Task marked complete but not implemented. Re-add to active?" **CRITICAL: Always verify against codebase, not just task status** ### Step 4: Show Parsed Tasks to User - Display parsed tasks via AskUserQuestion for approval - Show which file(s) they will be written to ### Step 5: Update Task Files After Approval **For single-file structure:** - Update tasks-active.json using Write or Edit tool **For multi-file structure:** - Update relevant \`tasks-active_{domain}.json\` file(s) - Update \`tasks-index.json\` with new task counts and domains Remember: Get user approval before writing any files. `.trim();