nodebench-mcp 2.11.0 → 2.13.0

package/dist/tools/boilerplateTools.js

@@ -54,52 +54,52 @@ function generatePackageJson(projectName, techStack) {
     }, null, 2);
 }
 function generateAgentsMd(projectName, techStack) {
-    return `# ${projectName} — Agent Instructions
-
-## Project Overview
-- **Name**: ${projectName}
-- **Tech Stack**: ${techStack}
-- **Created**: ${new Date().toISOString().split("T")[0]}
-
-## Quick Start for AI Agents
-
-### First Session
-1. Call \`bootstrap_project\` to register this project with NodeBench MCP
-2. Call \`search_all_knowledge\` to check for relevant past findings
-3. Call \`getMethodology("overview")\` to see all available methodologies
-
-### Every Task
-1. \`search_all_knowledge\` — Check what the system already knows
-2. \`start_verification_cycle\` — Begin 6-phase verification
-3. Follow phases 1-6 (guided by tool responses)
-4. \`run_mandatory_flywheel\` — 6-step final check before declaring done
-5. \`record_learning\` — Capture what you discovered
-
-### Progressive Tool Discovery
-- \`discover_tools("what you want to do")\` — Hybrid search with relevance scoring
-- \`get_tool_quick_ref("tool_name")\` — What to do after calling any tool
-- \`get_workflow_chain("new_feature")\` — Full step-by-step tool sequences
-
-## Architecture
-<!-- Describe your project architecture here -->
-
-## Conventions
-<!-- Describe coding conventions, patterns, and standards here -->
-
-## Quality Gates
-- All changes must pass \`run_mandatory_flywheel\` before shipping
-- UI changes must pass the \`ui_ux_qa\` quality gate
-- Code changes must pass \`run_closed_loop\` (compile→lint→test→debug)
-
-## Known Gotchas
-<!-- Record known issues and workarounds here. Also stored in NodeBench learnings DB. -->
-
-## Parallel Agent Coordination
-When using multiple agents (Claude Code subagents, worktrees, or terminals):
-1. \`claim_agent_task\` before starting work (prevents duplicate effort)
-2. \`assign_agent_role\` for specialization
-3. \`release_agent_task\` with progress note when done
-4. \`get_parallel_status\` to see all agent activity
+    return `# ${projectName} — Agent Instructions
+
+## Project Overview
+- **Name**: ${projectName}
+- **Tech Stack**: ${techStack}
+- **Created**: ${new Date().toISOString().split("T")[0]}
+
+## Quick Start for AI Agents
+
+### First Session
+1. Call \`bootstrap_project\` to register this project with NodeBench MCP
+2. Call \`search_all_knowledge\` to check for relevant past findings
+3. Call \`getMethodology("overview")\` to see all available methodologies
+
+### Every Task
+1. \`search_all_knowledge\` — Check what the system already knows
+2. \`start_verification_cycle\` — Begin 6-phase verification
+3. Follow phases 1-6 (guided by tool responses)
+4. \`run_mandatory_flywheel\` — 6-step final check before declaring done
+5. \`record_learning\` — Capture what you discovered
+
+### Progressive Tool Discovery
+- \`discover_tools("what you want to do")\` — Hybrid search with relevance scoring
+- \`get_tool_quick_ref("tool_name")\` — What to do after calling any tool
+- \`get_workflow_chain("new_feature")\` — Full step-by-step tool sequences
+
+## Architecture
+<!-- Describe your project architecture here -->
+
+## Conventions
+<!-- Describe coding conventions, patterns, and standards here -->
+
+## Quality Gates
+- All changes must pass \`run_mandatory_flywheel\` before shipping
+- UI changes must pass the \`ui_ux_qa\` quality gate
+- Code changes must pass \`run_closed_loop\` (compile→lint→test→debug)
+
+## Known Gotchas
+<!-- Record known issues and workarounds here. Also stored in NodeBench learnings DB. -->
+
+## Parallel Agent Coordination
+When using multiple agents (Claude Code subagents, worktrees, or terminals):
+1. \`claim_agent_task\` before starting work (prevents duplicate effort)
+2. \`assign_agent_role\` for specialization
+3. \`release_agent_task\` with progress note when done
+4. \`get_parallel_status\` to see all agent activity
 `;
 }
 function generateMcpConfig(projectName) {
@@ -114,116 +114,116 @@ function generateMcpConfig(projectName) {
     }, null, 2);
 }
 function generateGithubActions() {
-    return `name: NodeBench Quality Gate
-on:
-  pull_request:
-    branches: [main]
-  push:
-    branches: [main]
-
-jobs:
-  quality-gate:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-node@v4
-        with:
-          node-version: '22'
-      - run: npm ci
-      - run: npm run build
-      - run: npm run test
-      - run: npm run lint
+    return `name: NodeBench Quality Gate
+on:
+  pull_request:
+    branches: [main]
+  push:
+    branches: [main]
+
+jobs:
+  quality-gate:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: '22'
+      - run: npm ci
+      - run: npm run build
+      - run: npm run test
+      - run: npm run lint
 `;
 }
 function generateReadme(projectName, techStack) {
-    return `# ${projectName}
-
-Built with NodeBench MCP methodology for rigorous AI-assisted development.
-
-## Quick Start
-
-\`\`\`bash
-npm install
-npm run build
-npm run test
-\`\`\`
-
-## NodeBench MCP Integration
-
-This project is pre-configured for [NodeBench MCP](https://github.com/nodebench/nodebench-ai) — tools that make AI agents catch the bugs they normally ship.
-
-### For AI Agents
-See [AGENTS.md](./AGENTS.md) for detailed instructions.
-
-### Key Commands
-- \`npm run mcp:start\` — Start NodeBench MCP (full toolset)
-- \`npm run mcp:lite\` — Lightweight mode (34 tools)
-- \`npm run mcp:core\` — Core mode (79 tools)
-
-### MCP Configuration
-The \`.mcp.json\` file configures NodeBench MCP for your IDE.
-
-## Tech Stack
-${techStack}
-
-## Development Workflow
-1. **Research** → \`search_all_knowledge\`, \`run_recon\`
-2. **Implement** → Write code following conventions in AGENTS.md
-3. **Test** → \`run_closed_loop\` (compile→lint→test→debug)
-4. **Verify** → \`run_mandatory_flywheel\` (6-step verification)
-5. **Ship** → \`record_learning\`, \`promote_to_eval\`
+    return `# ${projectName}
+
+Built with NodeBench MCP methodology for rigorous AI-assisted development.
+
+## Quick Start
+
+\`\`\`bash
+npm install
+npm run build
+npm run test
+\`\`\`
+
+## NodeBench MCP Integration
+
+This project is pre-configured for [NodeBench MCP](https://github.com/nodebench/nodebench-ai) — tools that make AI agents catch the bugs they normally ship.
+
+### For AI Agents
+See [AGENTS.md](./AGENTS.md) for detailed instructions.
+
+### Key Commands
+- \`npm run mcp:start\` — Start NodeBench MCP (full toolset)
+- \`npm run mcp:lite\` — Lightweight mode (34 tools)
+- \`npm run mcp:core\` — Core mode (79 tools)
+
+### MCP Configuration
+The \`.mcp.json\` file configures NodeBench MCP for your IDE.
+
+## Tech Stack
+${techStack}
+
+## Development Workflow
+1. **Research** → \`search_all_knowledge\`, \`run_recon\`
+2. **Implement** → Write code following conventions in AGENTS.md
+3. **Test** → \`run_closed_loop\` (compile→lint→test→debug)
+4. **Verify** → \`run_mandatory_flywheel\` (6-step verification)
+5. **Ship** → \`record_learning\`, \`promote_to_eval\`
 `;
 }
 function generateParallelAgentsReadme() {
-    return `# Parallel Agent Coordination
-
-This directory supports multi-agent workflows with NodeBench MCP.
-
-## Files
-- \`current_tasks/\` — Lock files for claimed tasks
-- \`oracle/\` — Known-good reference outputs for oracle testing
-- \`roles.json\` — Agent role assignments
-- \`progress.md\` — Running status for agent orientation
-
-## Usage
-1. \`claim_agent_task({ taskKey: "...", description: "..." })\`
-2. Do the work
-3. \`release_agent_task({ taskKey: "...", status: "completed", progressNote: "..." })\`
-
-See AGENTS.md for full protocol.
+    return `# Parallel Agent Coordination
+
+This directory supports multi-agent workflows with NodeBench MCP.
+
+## Files
+- \`current_tasks/\` — Lock files for claimed tasks
+- \`oracle/\` — Known-good reference outputs for oracle testing
+- \`roles.json\` — Agent role assignments
+- \`progress.md\` — Running status for agent orientation
+
+## Usage
+1. \`claim_agent_task({ taskKey: "...", description: "..." })\`
+2. Do the work
+3. \`release_agent_task({ taskKey: "...", status: "completed", progressNote: "..." })\`
+
+See AGENTS.md for full protocol.
 `;
 }
 function generateProgressMd(projectName) {
-    return `# ${projectName} — Progress Tracker
-
-## Current Status
-- **Phase**: Setup
-- **Last Updated**: ${new Date().toISOString()}
-- **Active Agents**: 0
-
-## Completed Tasks
-<!-- Tasks will be logged here as agents complete them -->
-
-## Blocked Tasks
-<!-- Tasks that need fresh eyes or external input -->
-
-## Architecture Decisions
-<!-- Key decisions and their rationale -->
-
-## Known Issues
-<!-- Active issues to be aware of -->
+    return `# ${projectName} — Progress Tracker
+
+## Current Status
+- **Phase**: Setup
+- **Last Updated**: ${new Date().toISOString()}
+- **Active Agents**: 0
+
+## Completed Tasks
+<!-- Tasks will be logged here as agents complete them -->
+
+## Blocked Tasks
+<!-- Tasks that need fresh eyes or external input -->
+
+## Architecture Decisions
+<!-- Key decisions and their rationale -->
+
+## Known Issues
+<!-- Active issues to be aware of -->
 `;
 }
 function generateGitignore() {
-    return `node_modules/
-dist/
-.env
-.env.local
-*.log
-.nodebench/
-.parallel-agents/current_tasks/*.lock
-.tmp/
-coverage/
+    return `node_modules/
+dist/
+.env
+.env.local
+*.log
+.nodebench/
+.parallel-agents/current_tasks/*.lock
+.tmp/
+coverage/
 `;
 }
 function generateTsConfig() {

package/dist/tools/cCompilerBenchmarkTools.js

@@ -109,39 +109,39 @@ const CHALLENGES = {
 // ── DB schema extension ──────────────────────────────────────────────────
 function ensureBenchmarkTables() {
     const db = getDb();
-    db.exec(`
-    CREATE TABLE IF NOT EXISTS autonomy_benchmarks (
-      id TEXT PRIMARY KEY,
-      challenge_key TEXT NOT NULL,
-      challenge_name TEXT NOT NULL,
-      status TEXT NOT NULL DEFAULT 'active',
-      started_at TEXT NOT NULL DEFAULT (datetime('now')),
-      completed_at TEXT,
-      total_points INTEGER NOT NULL DEFAULT 0,
-      earned_points INTEGER NOT NULL DEFAULT 0,
-      milestones_completed INTEGER NOT NULL DEFAULT 0,
-      milestones_total INTEGER NOT NULL DEFAULT 0,
-      duration_minutes REAL,
-      tools_used TEXT,
-      context_tokens_estimate INTEGER,
-      notes TEXT,
-      score_pct REAL
-    );
-    CREATE TABLE IF NOT EXISTS benchmark_milestones (
-      id TEXT PRIMARY KEY,
-      benchmark_id TEXT NOT NULL,
-      milestone_id TEXT NOT NULL,
-      milestone_name TEXT NOT NULL,
-      status TEXT NOT NULL DEFAULT 'pending',
-      points INTEGER NOT NULL DEFAULT 0,
-      started_at TEXT,
-      completed_at TEXT,
-      duration_minutes REAL,
-      tools_used TEXT,
-      verification_passed INTEGER,
-      notes TEXT,
-      FOREIGN KEY (benchmark_id) REFERENCES autonomy_benchmarks(id)
-    );
+    db.exec(`
+    CREATE TABLE IF NOT EXISTS autonomy_benchmarks (
+      id TEXT PRIMARY KEY,
+      challenge_key TEXT NOT NULL,
+      challenge_name TEXT NOT NULL,
+      status TEXT NOT NULL DEFAULT 'active',
+      started_at TEXT NOT NULL DEFAULT (datetime('now')),
+      completed_at TEXT,
+      total_points INTEGER NOT NULL DEFAULT 0,
+      earned_points INTEGER NOT NULL DEFAULT 0,
+      milestones_completed INTEGER NOT NULL DEFAULT 0,
+      milestones_total INTEGER NOT NULL DEFAULT 0,
+      duration_minutes REAL,
+      tools_used TEXT,
+      context_tokens_estimate INTEGER,
+      notes TEXT,
+      score_pct REAL
+    );
+    CREATE TABLE IF NOT EXISTS benchmark_milestones (
+      id TEXT PRIMARY KEY,
+      benchmark_id TEXT NOT NULL,
+      milestone_id TEXT NOT NULL,
+      milestone_name TEXT NOT NULL,
+      status TEXT NOT NULL DEFAULT 'pending',
+      points INTEGER NOT NULL DEFAULT 0,
+      started_at TEXT,
+      completed_at TEXT,
+      duration_minutes REAL,
+      tools_used TEXT,
+      verification_passed INTEGER,
+      notes TEXT,
+      FOREIGN KEY (benchmark_id) REFERENCES autonomy_benchmarks(id)
+    );
   `);
 }
 // ── Tools ────────────────────────────────────────────────────────────────

package/dist/tools/documentationTools.js

@@ -24,65 +24,65 @@ async function canImport(pkg) {
 // ─── AGENTS.md template ───────────────────────────────────────────────────────
 const QUICK_REFS_PATTERN = "**→ Quick Refs:**";
 const QUICK_REFS_REGEX = /\*\*→ Quick Refs:\*\*/;
-const AGENTS_MD_TEMPLATE = `# AGENTS.md
-
-This file provides instructions for AI agents working on this project.
-
-**→ Quick Refs:** Start with [Project Overview](#project-overview) | Set up: [Development Setup](#development-setup) | Run: [Key Commands](#key-commands)
-
----
-
-## Project Overview
-
-<!-- Describe your project's purpose, architecture, and key components -->
-
-**→ Quick Refs:** Tech details: [Tech Stack](#tech-stack) | Code style: [Coding Conventions](#coding-conventions)
-
-## Development Setup
-
-<!-- Document setup steps: dependencies, environment variables, build commands -->
-
-**→ Quick Refs:** Key commands: [Key Commands](#key-commands) | Testing: [Testing](#testing)
-
-## Tech Stack
-
-<!-- List frameworks, libraries, and tools used -->
-
-**→ Quick Refs:** Conventions: [Coding Conventions](#coding-conventions) | Setup: [Development Setup](#development-setup)
-
-## Coding Conventions
-
-<!-- Document code style, naming conventions, patterns to follow -->
-
-**→ Quick Refs:** Testing: [Testing](#testing) | Edge cases: [Edge Cases & Learnings](#edge-cases--learnings)
-
-## Testing
-
-<!-- Document how to run tests, what testing frameworks are used -->
-
-**→ Quick Refs:** Commands: [Key Commands](#key-commands) | Edge cases: [Edge Cases & Learnings](#edge-cases--learnings)
-
-## Key Commands
-
-\`\`\`bash
-# Add your common commands here
-npm install
-npm run build
-npm run test
-\`\`\`
-
-**→ Quick Refs:** Setup: [Development Setup](#development-setup) | Testing: [Testing](#testing)
-
-## Edge Cases & Learnings
-
-<!-- Document gotchas, edge cases, and lessons learned -->
-
-**→ Quick Refs:** Conventions: [Coding Conventions](#coding-conventions) | Testing: [Testing](#testing)
-
----
-
-*This file is maintained by NodeBench MCP. Use \`update_agents_md\` to add learnings automatically.*
-*Every section MUST include \`**→ Quick Refs:**\` for agent chunking context.*
+const AGENTS_MD_TEMPLATE = `# AGENTS.md
+
+This file provides instructions for AI agents working on this project.
+
+**→ Quick Refs:** Start with [Project Overview](#project-overview) | Set up: [Development Setup](#development-setup) | Run: [Key Commands](#key-commands)
+
+---
+
+## Project Overview
+
+<!-- Describe your project's purpose, architecture, and key components -->
+
+**→ Quick Refs:** Tech details: [Tech Stack](#tech-stack) | Code style: [Coding Conventions](#coding-conventions)
+
+## Development Setup
+
+<!-- Document setup steps: dependencies, environment variables, build commands -->
+
+**→ Quick Refs:** Key commands: [Key Commands](#key-commands) | Testing: [Testing](#testing)
+
+## Tech Stack
+
+<!-- List frameworks, libraries, and tools used -->
+
+**→ Quick Refs:** Conventions: [Coding Conventions](#coding-conventions) | Setup: [Development Setup](#development-setup)
+
+## Coding Conventions
+
+<!-- Document code style, naming conventions, patterns to follow -->
+
+**→ Quick Refs:** Testing: [Testing](#testing) | Edge cases: [Edge Cases & Learnings](#edge-cases--learnings)
+
+## Testing
+
+<!-- Document how to run tests, what testing frameworks are used -->
+
+**→ Quick Refs:** Commands: [Key Commands](#key-commands) | Edge cases: [Edge Cases & Learnings](#edge-cases--learnings)
+
+## Key Commands
+
+\`\`\`bash
+# Add your common commands here
+npm install
+npm run build
+npm run test
+\`\`\`
+
+**→ Quick Refs:** Setup: [Development Setup](#development-setup) | Testing: [Testing](#testing)
+
+## Edge Cases & Learnings
+
+<!-- Document gotchas, edge cases, and lessons learned -->
+
+**→ Quick Refs:** Conventions: [Coding Conventions](#coding-conventions) | Testing: [Testing](#testing)
+
+---
+
+*This file is maintained by NodeBench MCP. Use \`update_agents_md\` to add learnings automatically.*
+*Every section MUST include \`**→ Quick Refs:**\` for agent chunking context.*
 `;
 function parseSections(content) {
     const lines = content.split("\n");

package/dist/tools/flywheelTools.js

@@ -208,12 +208,12 @@ export const flywheelTools = [
                 cycleId,
                 title,
                 linkedEvalRun: evalRunId,
-                phase1Instructions: `Phase 1: Context Gathering — Investigate the regression.
-The eval run "${evalRun.name}" showed regression. Research:
-- What changed since the baseline eval?
-- Which test cases failed that previously passed?
-- Is this a code change, upstream API change, or data drift?
-
+                phase1Instructions: `Phase 1: Context Gathering — Investigate the regression.
+The eval run "${evalRun.name}" showed regression. Research:
+- What changed since the baseline eval?
+- Which test cases failed that previously passed?
+- Is this a code change, upstream API change, or data drift?
+
 Start by calling search_learnings to check for known related issues.`,
             };
         },

package/dist/tools/learningTools.js

@@ -42,15 +42,15 @@ export const learningTools = [
             const db = getDb();
             const now = new Date().toISOString();
             const tagsJson = tags ? JSON.stringify(tags) : null;
-            db.prepare(`
-        INSERT INTO learnings (key, content, category, tags, source_cycle, created_at, updated_at)
-        VALUES (?, ?, ?, ?, ?, ?, ?)
-        ON CONFLICT(key) DO UPDATE SET
-          content = excluded.content,
-          category = excluded.category,
-          tags = excluded.tags,
-          source_cycle = excluded.source_cycle,
-          updated_at = excluded.updated_at
+            db.prepare(`
+        INSERT INTO learnings (key, content, category, tags, source_cycle, created_at, updated_at)
+        VALUES (?, ?, ?, ?, ?, ?, ?)
+        ON CONFLICT(key) DO UPDATE SET
+          content = excluded.content,
+          category = excluded.category,
+          tags = excluded.tags,
+          source_cycle = excluded.source_cycle,
+          updated_at = excluded.updated_at
       `).run(key, content, category, tagsJson, sourceCycle ?? null, now, now);
             const row = db
                 .prepare("SELECT id FROM learnings WHERE key = ?")
@@ -93,28 +93,28 @@ export const learningTools = [
             let results;
             if (args.category) {
                 results = db
-                    .prepare(`
-          SELECT l.id, l.key, l.content, l.category, l.tags, l.source_cycle, l.created_at,
-                 rank
-          FROM learnings_fts
-          JOIN learnings l ON l.id = learnings_fts.rowid
-          WHERE learnings_fts MATCH ?
-            AND l.category = ?
-          ORDER BY rank
-          LIMIT ?
+                    .prepare(`
+          SELECT l.id, l.key, l.content, l.category, l.tags, l.source_cycle, l.created_at,
+                 rank
+          FROM learnings_fts
+          JOIN learnings l ON l.id = learnings_fts.rowid
+          WHERE learnings_fts MATCH ?
+            AND l.category = ?
+          ORDER BY rank
+          LIMIT ?
         `)
                     .all(query, args.category, limit);
             }
             else {
                 results = db
-                    .prepare(`
-          SELECT l.id, l.key, l.content, l.category, l.tags, l.source_cycle, l.created_at,
-                 rank
-          FROM learnings_fts
-          JOIN learnings l ON l.id = learnings_fts.rowid
-          WHERE learnings_fts MATCH ?
-          ORDER BY rank
-          LIMIT ?
+                    .prepare(`
+          SELECT l.id, l.key, l.content, l.category, l.tags, l.source_cycle, l.created_at,
+                 rank
+          FROM learnings_fts
+          JOIN learnings l ON l.id = learnings_fts.rowid
+          WHERE learnings_fts MATCH ?
+          ORDER BY rank
+          LIMIT ?
         `)
                     .all(query, limit);
             }

package/dist/tools/localFileTools.d.ts

@@ -14,4 +14,7 @@
  * - tesseract.js: Image OCR (PNG/JPG/etc)
  */
 import type { McpTool } from "../types.js";
+/** General-purpose local file parsing tools (19 tools) */
 export declare const localFileTools: McpTool[];
+/** Specialized GAIA media image solver tools (6 tools) */
+export declare const gaiaMediaSolvers: McpTool[];