Skip to main content
Glama
MCP Engineering Agent

MCP Engineering Agent

by Lekha-2023

MCP Engineering Agent

A security-first engineering agent demonstrating an MCP-style JSON-RPC tool boundary. It separates model orchestration from typed tools, authorization, and audit events.

Architecture

User -> Agent -> JSON-RPC/MCP Boundary -> Tool Registry -> Authorization -> Tool
                                                    `-> Audit Event

Related MCP server: enterprise-agent-lab

What this demonstrates

  • tools/list and tools/call JSON-RPC methods

  • Typed tool registration and discovery

  • Role-based authorization before tool execution

  • Read-only vs write-capable tool separation

  • UTC audit events for every authorization decision

  • Framework-light design that is easy to extend to a full MCP server

Why it matters

Agentic systems become risky when an LLM can directly execute arbitrary actions. This project treats tools as a privileged interface with explicit schemas and authorization boundaries.

Run

pip install -e '.[dev]'
python -m app.server
pytest -q

No real credentials or external systems are required for the demo.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables controlled AI-agent access to enterprise-shaped tools with a deny-by-default gated write path, human approval, dry-run execution, and append-only audit logging.
      1
      -
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables LLM agents to safely invoke enterprise MCP tools under an enforceable guard that applies allowlist policies, caller roles, parameter restrictions, human confirmation for destructive operations, and a correlated JSONL audit log.
      43 PyPI
      AGPL 3.0
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables secure, cloud-agnostic MCP tool execution by authenticating agents, authorizing exact tools and arguments, enforcing independent one-time approvals, pinning upstream servers, and recording auditable decisions.
      MIT