devMCP

Modular Model Context Protocol (MCP) development server for local/offline coding agents, featuring a Telegram bot for remote management.

Overview

devMCP provides a suite of tools and workflows designed for autonomous coding agents. It follows a modular architecture where basic tools (CLI wrappers) are composed into higher-level "skills" (workflows).

Core Components

• server.py: The main entry point for the FastMCP server. It listens on 0.0.0.0:8000 using the streamable-http transport.
• telegram_bot.py: A standalone service providing two-way communication between the developer and agents via Telegram. It supports status checks, deployment, and delegation to agents like Kimi, Gemini, and Codex.
• registry.py: Manages the registration of all tool and skill modules into the MCP server.
• core/runtime.py: Shared utilities for command execution, file management, state (memory), and tool auto-installation.

Architecture

/
├── server.py             # FastMCP Entry Point
├── telegram_bot.py       # Telegram Interface
├── registry.py           # Module Registration
├── core/                 # Shared Runtime & Helpers
├── tools/                # Atomic Tool Wrappers (git, docker, files, etc.)
└── skills/               # Multi-step Workflow Tools

Setup & Run

1. Configure Environment

Copy the example environment file and fill in your credentials:

cp .env.example .env
# Edit .env and add your TELEGRAM_TOKEN and TELEGRAM_CHAT_ID

2. Prerequisites

• Python 3.10+
• Virtual environment (recommended)
• Tools like docker, git, ripgrep, etc. (the server will attempt to auto-install missing dev tools)

3. Running the MCP Server

/home/$user/mcp-dev-server/.venv/bin/python3 server.py

Running the Telegram Bot

/home/$user/mcp-dev-server/.venv/bin/python3 telegram_bot.py

Available Toolsets

Tools are categorized by their function in the tools/ directory:

• Filesystem: files.py (read, write, replace, list, search, glob, etc.)
• Git: git.py (status, diff, log, commit, push, pull)
• Docker: docker.py (logs, ps, restart, build, deploy, audit)
• Languages: python_dev.py, javascript_dev.py (linting, formatting, type checking)
• Quality & Security: quality.py, security.py (complexity, dead code, secrets, vulnerability scans)
• Infrastructure: database.py, network.py, remote.py (SQL/Redis queries, port scans, SSH management)
• Agents & Memory: agents.py, memory.py (delegation, persistent key-value store)

Skills (Workflows)

Higher-level tools found in skills/:

• ops_workflows.py: CI debugging, PR reviews, dependency upgrades, Docker optimization, incident triage.
• project_health.py: Broad project health checks, status reports.
• scaffold.py: Creating new tool and skill templates.
• workflows.py: Common multi-step tasks like release note generation.

Telegram Commands

The bot supports direct commands and agent delegation:

• deploy <project>: Local build and restart.
• status <project>: Docker compose status.
• logs <project> <service>: View service logs.
• git status/pull/push: Git operations.
• memory list/get: Access agent memory.
• kimi: <prompt>, gemini: <prompt>, codex: <prompt>: Delegate tasks to specific agents.
• <anything else>: Sent to Codex as a plain prompt.