# TRUSS MCP Server

**The structural layer for AI-native workflows.**

[![npm](https://img.shields.io/npm/v/truss-mcp)](https://www.npmjs.com/package/truss-mcp)
[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](./LICENSE)
[![MCP](https://img.shields.io/badge/MCP-compatible-brightgreen)](https://modelcontextprotocol.io)

AI-native developer tools for Claude Code. Generate CLAUDE.md files, rules, agent definitions, prompt templates, n8n workflows, security audits, and more — all from within your Claude Code session.

---

## What's Included

### Free Tier (no license needed)

| Tool | Description |
|------|-------------|
| `truss_init_project` | Initialize a project with CLAUDE.md, rules, and agent definitions |
| `truss_generate_claude_md` | Generate a CLAUDE.md from codebase analysis |
| `truss_search_prompts` | Search the prompt template library (3 results) |

### Pro Tier ($9/mo)

| Tool | Description |
|------|-------------|
| `truss_generate_rules` | Generate .claude/rules/ files for any language + framework |
| `truss_generate_agents` | Generate .claude/agents/ definitions for specialized AI agents |
| `truss_generate_workflow` | Generate n8n workflow JSON from natural language |
| `truss_optimize_config` | Audit and optimize your existing Claude Code configuration |
| `truss_prompt_library` | Full prompt template library (20+ templates, 6 categories) |
| `truss_security_audit` | Scan code for security vulnerabilities with remediation |
| `truss_test_generator` | Generate test cases from code (JS/TS, Python, Go, Rust) |

[Get TRUSS Pro](https://claw-factory.github.io/truss-site)

---

## Installation

```bash
npm install -g truss-mcp
```

## Configuration

Add to your Claude Code MCP configuration (`~/.claude/claude_desktop_config.json` or project-level):

```json
{
  "mcpServers": {
    "truss": {
      "command": "truss-mcp",
      "env": {
        "TRUSS_LICENSE_KEY": "truss_your_license_key_here"
      }
    }
  }
}
```

For the free tier, omit the `env` block:

```json
{
  "mcpServers": {
    "truss": {
      "command": "truss-mcp"
    }
  }
}
```

---

## Usage Examples

### Initialize a Project (Free)

> "Use truss_init_project to set up my Node.js API project"

Generates CLAUDE.md, code style rules, testing rules, and a QA reviewer agent definition. Supports: Node.js, Python, Go, Rust, React, fullstack, monorepo, and library project types.

### Generate CLAUDE.md (Free)

> "Use truss_generate_claude_md to create a CLAUDE.md for my project"

Analyzes your file structure and generates a tailored CLAUDE.md with tech stack detection, build commands, and code style guidelines.

### Search Prompts (Free)

> "Use truss_search_prompts to find prompts about security review"

Searches the built-in library of 20+ prompt templates across code review, testing, documentation, security, architecture, and debugging categories.

### Generate Rules (Pro)

> "Use truss_generate_rules for TypeScript with the React framework"

Creates comprehensive .claude/rules/ files with language-specific code style, testing, and security rules. Supports framework-specific additions (React, Express, FastAPI, Gin, Actix, Next.js, Django).

### Generate Agents (Pro)

> "Use truss_generate_agents to create a Security Auditor agent"

Generates .claude/agents/ definitions with pre-built templates for common roles (QA Reviewer, Security Auditor, Documentation Writer, Performance Analyst, API Designer).

### Generate Workflow (Pro)

> "Use truss_generate_workflow: When a GitHub issue is created, send a Slack notification"

Generates importable n8n workflow JSON with nodes, connections, and configuration. Auto-detects required nodes from your description.

### Optimize Config (Pro)

> "Use truss_optimize_config with my CLAUDE.md content"

Audits your configuration for completeness, structure, and effectiveness. Provides a grade (A-F), identifies missing sections, and suggests improvements.

### Security Audit (Pro)

> "Use truss_security_audit on these files"

Pattern-based security scanner that detects hardcoded secrets, SQL injection, XSS, command injection, path traversal, weak cryptography, SSRF, and more.

### Test Generator (Pro)

> "Use truss_test_generator for this Python function"

Generates test files with happy path, edge case, and error handling tests. Supports Vitest/Jest (JS/TS), pytest (Python), Go testing, and Rust built-in tests.

---

## Prompt Library Categories

The built-in prompt library includes templates for:

- **Code Review**: Security review, performance review, readability review, PR review
- **Testing**: Unit tests, integration tests, E2E scenarios
- **Documentation**: API docs, README generation, ADRs, changelog entries
- **Security**: Threat modeling, audit checklists, incident response playbooks
- **Architecture**: System design, migration plans, API design review
- **Debugging**: Root cause analysis, performance investigation, log analysis

---

## Template Coverage

CLAUDE.md templates for 8 project types:

- `node-api` — Node.js REST API (Express/Fastify)
- `python-ml` — Python ML/data science project
- `python-api` — Python API (FastAPI)
- `go-service` — Go microservice
- `rust-cli` — Rust CLI application
- `react-app` — React SPA (TypeScript + Vite)
- `fullstack` — Full-stack application (frontend + backend + DB)
- `monorepo` — Monorepo (Turborepo/Nx)
- `library` — Publishable library/package

Rule templates for 4 languages: JavaScript/TypeScript, Python, Go, Rust

---

## Running Evals

```bash
cd truss-mcp-server
npm run eval
```

The eval suite tests all free and Pro tools for correctness, output structure, and edge cases.

---

## License

MIT -- see [LICENSE](./LICENSE).

---

Built by [TRUSS](https://claw-factory.github.io/truss-site) -- the structural layer for AI-native workflows.