melra

XAGI-Lab/melra
★ 13 stars TypeScript Security Updated 4d ago ✓ Clean
Open-source local MCP runtime that checks AI-agent tool calls, gates risky actions on approval, verifies results, and avoids blind retries after timeouts.
View on GitHub → 🛡️ Security & TRiSM Audit →

⚡ What This Server Does & Key Capabilities

The melra Model Context Protocol (MCP) server provides seamless integration between AI assistants (such as Claude Desktop, Cursor, and Claude Code) and Open-source local MCP runtime that checks AI-agent tool calls, gates risky actions on approval, verifies results, and avoids blind retries after timeouts. By connecting this server, your AI agent gains specialized capabilities to inspect, query, and automate tasks within the Security domain.

Core Agent Capabilities

💡 Example Prompts to Ask Your AI Agent

Once installed in Claude Desktop or Cursor, try asking your assistant:

"Scan the connected repository for hardcoded API keys or vulnerable dependencies."
"Audit the permissions of all connected MCP tools and flag potential injection risks."

Compatible AI Clients

🟣 Claude Desktop ⚡ Cursor IDE 💻 Claude Code CLI 🌊 Windsurf 🤖 Gemini CLI

Quick Install

Copy the config for your editor. Verify environment variables or setup notes in the README.

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "melra": {
      "command": "npx",
      "args": [
        "-y",
        "XAGI-Lab/melra"
      ]
    }
  }
}
⚡
Building Production MCP Agents? PRO TOOLKITS
Get the Complete 31,400+ MCP Ecosystem Dataset ($29) or the FastMCP + TypeScript Starter Kit ($39) with turnkey Docker & Cloudflare deployment.

README Documentation

<div align="center">


<img src="docs/assets/melra-logo.png" alt="MELRA logo" width="132" />


# MELRA


Modular Execution Layer for Reliable Autonomy


Your AI agent decides what to do. MELRA governs how tool calls are authorized, verified, and recovered.


An open-source local execution runtime for agent tool calls you route through MELRA.

It checks each call against policy, requests approval when needed, verifies declared

evidence, and keeps a redacted receipt of what runs to an outcome. For each task, MELRA

dispatches mutations at most once; uncertain outcomes are parked for reconciliation

instead of replay. Exactly-once external effects still depend on the provider. Calls

made through other tools remain outside MELRA's control.


<br />


<!-- Build & quality (live) -

View Full Documentation & Updates on GitHub ↗

Topics

securityruntimeguardrailstool-validationmcp-server