Mcp Devtools

io.github.marin1321Published by GitHub account marin1321v1.0.1Updated Oct 24, 2026

MCP server giving AI agents safe access to filesystem, databases, processes, and APIs.

VerifiedSTDIODesktop onlyDatabasesFiles & StorageDeveloper Tools

Overview

AI-generated overview

Local MCP server that gives an AI assistant scoped access to your dev environment: files, databases, processes and OpenAPI-based APIs.

What you can do
  • Read, write, list, search and inspect files inside a configured scope
  • Query databases, list tables and describe table schemas
  • Run allowlisted commands, read logs, list processes and read environment variables
  • Parse OpenAPI specs and call APIs restricted to the spec's declared hosts
  • Expose read-only resources (tool catalog, server info) and prompt templates for debugging, code review, codebase exploration and refactoring
  • Load custom tools through a plugin API
When to use it
Useful when you want an assistant to work directly against a local project: inspecting and editing files, querying a development database, running approved commands, or exercising an API described by an OpenAPI spec, with scope limits and an audit log instead of unrestricted shell access.
Requirements
Runs locally over stdio via npx (Node.js required) or over HTTP on a configurable port. Configuration comes from mcp-devtools.json, .mcp-devtoolsrc or a package.json key; zero-config defaults are supported. HTTP mode can require a bearer token supplied through an environment variable such as MCP_AUTH_TOKEN. Database access needs connection settings in the config.
Before you install
It can write files, run commands and call external APIs, so scope, command allowlist and database read-only mode should be configured before use. get_env exposes environment variables, with masking of secret-like names only. HTTP mode without a configured token is unauthenticated. Audit logging is opt-in, and plugins run arbitrary code at startup.

Installation

In SourceWeft

  1. Open Mcp Devtools in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

mcp-devtools

[npm version] [CI] [License: MIT] [Node.js]

AI-native developer tools via Model Context Protocol. A production-grade MCP server that gives AI agents (Claude, Cursor, Copilot, Continue, ...) safe, scoped access to your local development environment.

Why

The MCP ecosystem is full of single-purpose tutorials and vendor-locked adapters. There is no well-maintained, multi-tool, framework-agnostic, production-quality MCP package for everyday developer tooling.

mcp-devtools fills that gap with 14 tools, 3 MCP Resources, 4 MCP Prompts, a Plugin API, two transport modes (stdio + HTTP with auth), and an audit log — built on patterns refined in production at DailyBot.

Quick start

stdio (default)

bash
npx @oscarmarin/mcp-devtools

Add it to Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json):

json
{  "mcpServers": {    "devtools": {      "command": "npx",      "args": ["-y", "@oscarmarin/mcp-devtools"]    }  }}

Or Cursor (~/.cursor/mcp.json): same block.

HTTP transport

Create a mcp-devtools.json in your project root:

json
{  "transport": "http",  "port": 3333,  "auth": {    "token": "env:MCP_AUTH_TOKEN"  }}

Then start the server:

bash
npx @oscarmarin/mcp-devtools

The MCP endpoint will be available at http://localhost:3333.

Tools

GroupTools
Filesystemread_file, write_file, list_directory, search_files, get_file_info
Databasequery_db, list_tables, describe_table
Processrun_command, read_logs, get_env, list_processes
OpenAPIparse_openapi, call_api

Per-tool reference: docs/tools/.

MCP Resources

The server exposes read-only data via MCP Resources:

URIDescription
devtools://toolsCatalog of all registered tools with schemas
devtools://server-infoServer version, transport, scope, tool count

MCP Prompts

Curated prompt templates for common development workflows:

PromptDescription
debug_errorSystematically debug an error using mcp-devtools tools
code_reviewReview a file for bugs, security issues, and code quality
explore_codebaseExplore and understand a project's structure and conventions
refactor_functionRefactor a function for readability, performance, or testability

Plugin API

Extend mcp-devtools with custom tools — no fork required.

Config-based (load at startup):

json
{  "plugins": ["./my-tools.js", "@scope/mcp-plugin-foo"]}

Each plugin module default-exports an array of tool definitions:

typescript
import { defineTool } from "@oscarmarin/mcp-devtools";import { z } from "zod";
export default [  defineTool({    name: "my_tool",    description: "Does something useful",    inputSchema: z.object({ input: z.string() }),    handler: async (args, config) => ({      ok: true,      data: { result: args.input.toUpperCase() },    }),  }),];

Configuration

Configuration is loaded by cosmiconfig from mcp-devtools.json, .mcp-devtoolsrc, or the mcpDevtools key in package.json. See mcp-devtools.example.json and docs/configuration.md for the full schema.

Zero-config is supported: running npx @oscarmarin/mcp-devtools with no config uses schema defaults (RNF-05).

Security

Four non-bypassable controls:

  1. Filesystem scope boundary. Every path is resolved to an absolute and compared against config.scope. Symlinks that escape scope throw SCOPE_VIOLATION.
  2. Command allowlist. run_command only executes binaries whose basename is in allowedCommands. Invocation uses spawn(file, args) (no shell), so shell-injection via the command argument is structurally impossible.
  3. Database read-only mode. When readOnly: true, all SQL is parsed and INSERT/UPDATE/DELETE/DROP/CREATE/GRANT are rejected. Queries run in BEGIN READ ONLY ... ROLLBACK on PostgreSQL.
  4. HTTP Bearer auth. When auth.token is configured, every HTTP request must include Authorization: Bearer <token>. Comparison uses crypto.timingSafeEqual to prevent timing attacks.

Additional safety measures:

  • Audit log. Opt-in NDJSON log of every tool invocation with timing, sanitized inputs, and result status.
  • Secret masking. get_env automatically masks values matching common secret patterns (SECRET, TOKEN, PASSWORD, KEY, etc.).
  • OpenAPI host restriction. call_api only sends requests to hosts listed in the spec's servers array.
  • Output capping. All tools cap their output to prevent context flooding (100KB for commands, 1MB for files, 200 rows for queries).

Contributing

bash
git clone https://github.com/marin1321/mcp-devtools.gitcd mcp-devtoolsnpm installnpm run dev       # tsup --watchnpm run test      # vitestnpm run typecheck # tsc --noEmitnpm run lint      # eslint .

See CONTRIBUTING.md for the full workflow and CODE_OF_CONDUCT.md for community guidelines.

License

MIT © Oscar Humberto Marin Molina — oscarmarindev.com

Source: README.md at commit 5d4e8d6

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v1.0.1LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts