refactor: migrate plugin structure from .claude-plugin to project root

Restructure plugin to follow Claude Code official documentation: - Move TypeScript files from .claude-plugin/* to project root - Create Markdown command files in commands/ - Update plugin.json to reference ./commands/*.md - Add comprehensive plugin installation guide Changes: - Commands: pm.md, research.md, index-repo.md (new Markdown format) - TypeScript: pm/, research/, index/ moved to root - Hooks: hooks/hooks.json moved to root - Documentation: PLUGIN_INSTALL.md, updated CLAUDE.md, Makefile Note: This commit represents transition state. Original TypeScript-based execution system was replaced with Markdown commands. Further redesign needed to properly integrate Skills and Hooks per official docs. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
2025-12-29 16:16:08 +00:00 · 2025-10-21 16:37:35 +09:00
parent fbc74e4670
commit c91a3a4805
16 changed files with 1795 additions and 331 deletions
--- a/pm/confidence.ts
+++ b/pm/confidence.ts
@@ -0,0 +1,171 @@
+/**
+ * Confidence Check - Pre-implementation confidence assessment
+ *
+ * Prevents wrong-direction execution by assessing confidence BEFORE starting.
+ * Requires ≥90% confidence to proceed with implementation.
+ *
+ * Test Results (2025-10-21):
+ * - Precision: 1.000 (no false positives)
+ * - Recall: 1.000 (no false negatives)
+ * - 8/8 test cases passed
+ */
+
+export interface Context {
+  task?: string;
+  duplicate_check_complete?: boolean;
+  architecture_check_complete?: boolean;
+  official_docs_verified?: boolean;
+  oss_reference_complete?: boolean;
+  root_cause_identified?: boolean;
+  confidence_checks?: string[];
+  [key: string]: any;
+}
+
+/**
+ * Assess confidence level (0.0 - 1.0)
+ *
+ * Investigation Phase Checks:
+ * 1. No duplicate implementations? (25%)
+ * 2. Architecture compliance? (25%)
+ * 3. Official documentation verified? (20%)
+ * 4. Working OSS implementations referenced? (15%)
+ * 5. Root cause identified? (15%)
+ *
+ * @param context - Task context with investigation flags
+ * @returns Confidence score (0.0 = no confidence, 1.0 = absolute certainty)
+ */
+export async function confidenceCheck(context: Context): Promise<number> {
+  let score = 0.0;
+  const checks: string[] = [];
+
+  // Check 1: No duplicate implementations (25%)
+  if (noDuplicates(context)) {
+    score += 0.25;
+    checks.push("✅ No duplicate implementations found");
+  } else {
+    checks.push("❌ Check for existing implementations first");
+  }
+
+  // Check 2: Architecture compliance (25%)
+  if (architectureCompliant(context)) {
+    score += 0.25;
+    checks.push("✅ Uses existing tech stack (e.g., Supabase)");
+  } else {
+    checks.push("❌ Verify architecture compliance (avoid reinventing)");
+  }
+
+  // Check 3: Official documentation verified (20%)
+  if (hasOfficialDocs(context)) {
+    score += 0.2;
+    checks.push("✅ Official documentation verified");
+  } else {
+    checks.push("❌ Read official docs first");
+  }
+
+  // Check 4: Working OSS implementations referenced (15%)
+  if (hasOssReference(context)) {
+    score += 0.15;
+    checks.push("✅ Working OSS implementation found");
+  } else {
+    checks.push("❌ Search for OSS implementations");
+  }
+
+  // Check 5: Root cause identified (15%)
+  if (rootCauseIdentified(context)) {
+    score += 0.15;
+    checks.push("✅ Root cause identified");
+  } else {
+    checks.push("❌ Continue investigation to identify root cause");
+  }
+
+  // Store check results
+  context.confidence_checks = checks;
+
+  // Display checks
+  console.log("📋 Confidence Checks:");
+  checks.forEach(check => console.log(`   ${check}`));
+  console.log("");
+
+  return score;
+}
+
+/**
+ * Check for duplicate implementations
+ *
+ * Before implementing, verify:
+ * - No existing similar functions/modules (Glob/Grep)
+ * - No helper functions that solve the same problem
+ * - No libraries that provide this functionality
+ */
+function noDuplicates(context: Context): boolean {
+  return context.duplicate_check_complete ?? false;
+}
+
+/**
+ * Check architecture compliance
+ *
+ * Verify solution uses existing tech stack:
+ * - Supabase project → Use Supabase APIs (not custom API)
+ * - Next.js project → Use Next.js patterns (not custom routing)
+ * - Turborepo → Use workspace patterns (not manual scripts)
+ */
+function architectureCompliant(context: Context): boolean {
+  return context.architecture_check_complete ?? false;
+}
+
+/**
+ * Check if official documentation verified
+ *
+ * For testing: uses context flag 'official_docs_verified'
+ * For production: checks for README.md, CLAUDE.md, docs/ directory
+ */
+function hasOfficialDocs(context: Context): boolean {
+  // Check context flag (for testing and runtime)
+  if ('official_docs_verified' in context) {
+    return context.official_docs_verified ?? false;
+  }
+
+  // Fallback: check for documentation files (production)
+  // This would require filesystem access in Node.js
+  return false;
+}
+
+/**
+ * Check if working OSS implementations referenced
+ *
+ * Search for:
+ * - Similar open-source solutions
+ * - Reference implementations in popular projects
+ * - Community best practices
+ */
+function hasOssReference(context: Context): boolean {
+  return context.oss_reference_complete ?? false;
+}
+
+/**
+ * Check if root cause is identified with high certainty
+ *
+ * Verify:
+ * - Problem source pinpointed (not guessing)
+ * - Solution addresses root cause (not symptoms)
+ * - Fix verified against official docs/OSS patterns
+ */
+function rootCauseIdentified(context: Context): boolean {
+  return context.root_cause_identified ?? false;
+}
+
+/**
+ * Get recommended action based on confidence level
+ *
+ * @param confidence - Confidence score (0.0 - 1.0)
+ * @returns Recommended action
+ */
+export function getRecommendation(confidence: number): string {
+  if (confidence >= 0.9) {
+    return "✅ High confidence (≥90%) - Proceed with implementation";
+  } else if (confidence >= 0.7) {
+    return "⚠️ Medium confidence (70-89%) - Continue investigation, DO NOT implement yet";
+  } else {
+    return "❌ Low confidence (<70%) - STOP and continue investigation loop";
+  }
+}
--- a/pm/index.ts
+++ b/pm/index.ts
@@ -0,0 +1,159 @@
+/**
+ * PM Agent - Project Manager with Confidence-Driven Workflow
+ *
+ * Auto-executes on session start via hooks/hooks.json
+ * Orchestrates sub-agents with 90% confidence threshold
+ */
+
+import { execSync } from 'child_process';
+import { confidenceCheck } from './confidence';
+
+interface SessionContext {
+  gitStatus: string;
+  tokenBudget: number;
+  projectRoot: string;
+}
+
+/**
+ * Session Start Protocol
+ * Auto-executes when Claude Code starts
+ */
+export async function sessionStart(): Promise<void> {
+  console.log("🚀 PM Agent activated");
+
+  // 1. Check git status
+  const gitStatus = checkGitStatus();
+  console.log(`📊 Git: ${gitStatus}`);
+
+  // 2. Token budget check (from Claude Code UI)
+  console.log("💡 Check token budget with /context");
+
+  // 3. Ready
+  console.log("✅ PM Agent ready to accept tasks");
+  console.log("");
+  console.log("**Core Capabilities**:");
+  console.log("- 🔍 Pre-implementation confidence check (≥90% required)");
+  console.log("- ⚡ Parallel investigation and execution");
+  console.log("- 📊 Token-budget-aware operations");
+  console.log("");
+  console.log("**Usage**: Assign tasks directly - PM Agent will orchestrate");
+}
+
+/**
+ * Check git repository status
+ */
+function checkGitStatus(): string {
+  try {
+    const status = execSync('git status --porcelain', { encoding: 'utf-8' });
+    if (!status.trim()) {
+      return 'clean';
+    }
+    const lines = status.trim().split('\n').length;
+    return `${lines} file(s) modified`;
+  } catch {
+    return 'not a git repo';
+  }
+}
+
+/**
+ * Main task handler
+ * Called when user assigns a task
+ */
+export async function handleTask(task: string): Promise<void> {
+  console.log(`📝 Task received: ${task}`);
+  console.log("");
+
+  // Start confidence-driven workflow
+  await confidenceDrivenWorkflow(task);
+}
+
+/**
+ * Confidence-Driven Workflow
+ *
+ * 1. Investigation phase (loop until 90% confident)
+ * 2. Confidence check
+ * 3. Implementation (only when ≥90%)
+ */
+async function confidenceDrivenWorkflow(task: string): Promise<void> {
+  let confidence = 0;
+  let iteration = 0;
+  const MAX_ITERATIONS = 10;
+
+  console.log("🔍 Starting investigation phase...");
+  console.log("");
+
+  while (confidence < 0.9 && iteration < MAX_ITERATIONS) {
+    iteration++;
+    console.log(`🔄 Investigation iteration ${iteration}...`);
+
+    // Investigation actions (delegated to sub-agents)
+    const context = await investigate(task);
+
+    // Self-evaluate confidence
+    confidence = await confidenceCheck(context);
+
+    console.log(`📊 Confidence: ${(confidence * 100).toFixed(0)}%`);
+
+    if (confidence < 0.9) {
+      console.log("⚠️ Confidence < 90% - Continue investigation");
+      console.log("");
+    }
+  }
+
+  if (confidence >= 0.9) {
+    console.log("✅ High confidence (≥90%) - Proceeding to implementation");
+    console.log("");
+    // Implementation phase
+    await implement(task);
+  } else {
+    console.log("❌ Max iterations reached - Request user clarification");
+  }
+}
+
+/**
+ * Investigation phase
+ * Delegates to sub-agents: research, index, grep, etc.
+ */
+async function investigate(task: string): Promise<any> {
+  // This will be orchestrated by Claude using:
+  // - /research for web research
+  // - /index-repo for codebase structure
+  // - Glob/Grep for code search
+  // - WebFetch for official docs
+
+  return {
+    task,
+    duplicate_check_complete: false,
+    architecture_check_complete: false,
+    official_docs_verified: false,
+    oss_reference_complete: false,
+    root_cause_identified: false
+  };
+}
+
+/**
+ * Implementation phase
+ * Only executed when confidence ≥ 90%
+ */
+async function implement(task: string): Promise<void> {
+  console.log(`🚀 Implementing: ${task}`);
+  // Actual implementation delegated to Claude
+}
+
+/**
+ * Memory Management (Mindbase MCP integration)
+ * Zero-footprint: No auto-load, explicit load/save only
+ */
+export const memory = {
+  load: async () => {
+    console.log("💾 Use /sc:load to load context from Mindbase MCP");
+  },
+  save: async () => {
+    console.log("💾 Use /sc:save to persist session to Mindbase MCP");
+  }
+};
+
+// Auto-execute on session start
+if (require.main === module) {
+  sessionStart();
+}
--- a/pm/package.json
+++ b/pm/package.json
@@ -0,0 +1,18 @@
+{
+  "name": "@pm-agent/core",
+  "version": "1.0.0",
+  "description": "PM Agent - Project Manager with 90% confidence checks",
+  "main": "index.ts",
+  "scripts": {
+    "test": "jest",
+    "build": "tsc"
+  },
+  "dependencies": {},
+  "devDependencies": {
+    "@types/node": "^20.0.0",
+    "typescript": "^5.0.0"
+  },
+  "engines": {
+    "node": ">=18.0.0"
+  }
+}