<div align="center">

# AgentSkillMesh

### High-Performance Model Context Protocol (MCP) Tool Suite & Skill Runtime
**Optimized for OpenAI Codex, ChatGPT Pro Workspace Agents, and Autonomous Coding Harnesses**

[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
[![Node.js](https://img.shields.io/badge/Node.js-%3E%3D18.0.0-brightgreen.svg)](https://nodejs.org/)
[![MCP](https://img.shields.io/badge/MCP-Standard%202.0-orange.svg)](https://modelcontextprotocol.io/)

</div>

---

## Overview

**AgentSkillMesh** provides foundational tools for AI coding agents communicating via the **Model Context Protocol (MCP)**. It equips **OpenAI Codex**, **ChatGPT Pro**, and coding harnesses with fast security static analysis, AST symbol indexing, and safe execution contexts.

## Key Capabilities

- Static Security Rule Engine: Instant linting for dynamic eval, secret leaks, and insecure regexes.
- AST Symbol Grep: Sub-millisecond symbol search across large repositories without embedding overhead.
- Zero External Bloat: Zero-dependency, pure ESM architecture designed for high-concurrency agent workflows.

## Tools Exposed

AgentSkillMesh exposes two MCP tools that any MCP-compatible agent (OpenAI Codex, Claude, etc.) can call:

| Tool | What it does |
|------|--------------|
| `scan_code_security` | Scans source code for security hazards: dynamic `eval`/`exec` usage (SEC-001), catastrophic-backtracking regexes (SEC-002), and leaked secret tokens — AWS keys, GitHub tokens, OpenAI keys (SEC-003). |
| `search_ast_symbols` | Fast AST-based symbol search across code — find definitions and references without embedding overhead. |

### Demo

```bash
$ node dist/cli.js scan ./vuln_demo.js
Scan completed for ./vuln_demo.js. Findings: 2
- [CRITICAL] Line 1: High-entropy secret token literal detected. (SEC-003)
- [CRITICAL] Line 3: Dangerous dynamic code evaluation construct detected. (SEC-001)
```

## Quick Start

```bash
# Clone and build
git clone https://github.com/miriyaladhanwinn/agent-skill-mesh.git
cd agent-skill-mesh
npm install
npm run build

# Start MCP stdio server
node dist/cli.js serve

# Scan a source file
node dist/cli.js scan ./path/to/file.js
```

## MCP Configuration

```json
{
  "mcpServers": {
    "agent-mesh": {
      "command": "node",
      "args": ["/path/to/agent-skill-mesh/dist/cli.js", "serve"]
    }
  }
}
```

## License

Licensed under the [Apache License, Version 2.0](LICENSE).
Maintained by [@miriyaladhanwinn](https://github.com/miriyaladhanwinn).