mcp-devenv

mcp-devenv

An MCP server that gives your AI assistant full awareness of your local dev environment — running processes, Docker containers, git state, open ports, log files, and more.

Category
Visit Server

README

iwomm-mcp

CI

An MCP server that gives your AI assistant full awareness of your local dev environment — running processes, Docker containers, git state, open ports, log files, and more.

Stop copy-pasting context. Let the AI ask for what it needs.

demo


Tools

Tool What it answers
get_running_processes What Node/Python/Go processes are running? What's on port 3000?
get_docker_state Which containers are up? Are they healthy? What are their ports?
get_git_status What branch am I on? What's dirty? Am I ahead/behind?
get_env_summary What keys does my .env have? (Values masked by default)
get_config_summary What's in my application.yml? (.yml, .properties, .env)
get_dependencies What libraries am I using? (pom.xml, build.gradle, go.mod, package.json)
get_open_ports What's listening on this machine right now?
get_recent_logs What did my API service log in the last 50 lines?
get_project_structure What does this codebase look like? (gitignore-aware)
check_service_health Is http://localhost:3000/health responding? How fast?

Language support

Most tools operate at the OS or network level and work regardless of stack. The table below calls out the gaps.

Tool Node / TS Java (Spring) Go Python Ruby / .NET / Rust
get_project_structure ✅ ✅ ✅ ✅ ✅
get_git_status ✅ ✅ ✅ ✅ ✅
get_docker_state ✅ ✅ ✅ ✅ ✅
get_open_ports ✅ ✅ ✅ ✅ ✅
check_service_health ✅ ✅ ✅ ✅ ✅
get_recent_logs ✅ ✅ ✅ ✅ ✅
get_running_processes ✅ ⚠️ all JVM processes share the name java — use filter_name=java + filter_args=myapp.jar ✅ ✅ ✅
get_env_summary ✅ ⚠️ .env files only — use get_config_summary for application.* ✅ ✅ ✅
get_config_summary ✅ ✅ .yml · .properties ❌ ❌ ❌
get_dependencies ✅ package.json ✅ pom.xml · build.gradle[.kts] ✅ go.mod ❌ ❌

What still needs copy-paste

Some things Claude can't read yet — paste these into chat when needed:

  • Python deps: requirements.txt, pyproject.toml, setup.py
  • Rust deps: Cargo.toml
  • Ruby deps: Gemfile
  • .NET deps: *.csproj
  • Python/Rust config: config.toml, settings.py and similar formats
  • Java stack traces in logs: get_recent_logs returns raw lines — a trace split across the tail boundary may arrive incomplete
  • Spring OAuth2 properties: keys containing auth (e.g. authorization-grant-type) are masked by the secret-detection heuristic; paste the value if needed

Installation

Via npm (recommended)

npm install -g iwomm-mcp

Or use it without installing — npx will fetch and run it on demand (see Claude config below).

From source

git clone https://github.com/dicoy/iwomm-mcp.git
cd iwomm-mcp
npm install
npm run build

Add to Claude Code

claude mcp add devenv -- npx -y iwomm-mcp

Add to Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json on macOS:

{
  "mcpServers": {
    "devenv": {
      "command": "npx",
      "args": ["-y", "iwomm-mcp"]
    }
  }
}

Restart Claude. You should see devenv in /mcp.


Optional: .mcp-context.yml

Drop this file at the root of any project to tell iwomm-mcp about your services:

# .mcp-context.yml
services:
  api:
    port: 3000
    logs: ./logs/api.log
    health_endpoint: /health
  worker:
    logs: ./logs/worker.log
  postgres:
    port: 5432

env_files:
  - .env
  - .env.local

# Keys matching these patterns will have their values revealed
# (even if they look like secrets). Supports exact match and * prefix glob.
reveal_env_patterns:
  - NODE_ENV
  - APP_*
  - FEATURE_*

Without this file, get_recent_logs works only with explicit file paths. All other tools work without any config.


Architecture

┌─────────────────────────────────────────────────────────┐
│                     Claude (AI)                          │
└───────────────────────┬─────────────────────────────────┘
                        │ MCP protocol (stdio)
┌───────────────────────▼─────────────────────────────────┐
│                    iwomm-mcp                             │
│                                                          │
│  ┌──────────────────┐     ┌──────────────────────────┐  │
│  │   Tool Registry  │     │       Providers          │  │
│  │                  │     │                          │  │
│  │  get_git_status ─┼────▶│  IGitProvider            │  │
│  │  get_docker_*   ─┼────▶│  IDockerProvider         │  │
│  │  get_processes  ─┼────▶│  IProcessProvider        │  │
│  │  ...             │     │  IEnvProvider            │  │
│  └──────────────────┘     │  ILogProvider            │  │
│                            │  IPortProvider           │  │
│  ┌──────────────────┐     └──────────────┬───────────┘  │
│  │  .mcp-context    │                     │              │
│  │     .yml         │           concrete impls           │
│  └──────────────────┘                     │              │
└───────────────────────────────────────────┼─────────────┘
                                            │
              ┌─────────────┬───────────────┼──────────────┐
              ▼             ▼               ▼              ▼
           ps/lsof      simple-git      dockerode       fs/readline

Design principles:

  • Provider pattern — every tool depends on an interface, not a concrete implementation. Tests mock the interface; the OS is never touched in tests.
  • One Zod schema per tool — the same schema drives both MCP input validation and TypeScript types. No duplication.
  • Typed error hierarchy — DockerNotAvailableError, GitNotARepositoryError, etc. Callers can catch exactly what they expect.
  • Composition root — createServer() in server.ts is the only place that wires interfaces to implementations. Everything else is pure.

Development

npm run dev          # build in watch mode
npm run typecheck    # tsc --noEmit
npm run lint         # biome check
npm run lint:fix     # biome check --write
npm run test         # vitest run
npm run test:watch   # vitest (interactive)
npm run ci           # typecheck + lint + test + build

Adding a new tool

  1. Create src/tools/your-tool-name/schema.ts — Zod input schema
  2. Create src/tools/your-tool-name/handler.ts — pure function, injected providers
  3. Create src/tools/your-tool-name/handler.test.ts — mock the providers, not the OS
  4. Register in src/registry/tool-registry.ts — one server.tool(...) call

If the tool needs a new system capability, add an interface to src/providers/ with a corresponding implementation. The handler never imports a concrete provider class.

Project structure

src/
├── config/
│   ├── loader.ts          # .mcp-context.yml parser
│   └── schema.ts          # Zod schema + inferred types
├── errors/
│   └── index.ts           # typed error hierarchy
├── providers/
│   ├── docker.ts          # IDockerProvider + DockerodeProvider
│   ├── env.ts             # IEnvProvider + FsEnvProvider
│   ├── git.ts             # IGitProvider + SimpleGitProvider
│   ├── log.ts             # ILogProvider + FsLogProvider
│   ├── port.ts            # IPortProvider + LsofPortProvider
│   └── process.ts         # IProcessProvider + NodeProcessProvider
├── registry/
│   └── tool-registry.ts   # wires tools to the MCP server
├── tools/
│   └── <tool-name>/
│       ├── schema.ts
│       ├── handler.ts
│       └── handler.test.ts
├── index.ts               # stdio transport + startup
└── server.ts              # createServer() factory

Tech stack

Runtime Node.js 20+
MCP SDK @modelcontextprotocol/sdk
Validation zod
Git simple-git
Docker dockerode
Shell execa
Build tsup
Tests vitest
Lint + format biome

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured