ROS 2 Workspace Inspector MCP

ROS 2 Workspace Inspector MCP

Provides static analysis of ROS 2 workspaces, enabling inspection of packages, dependencies, interfaces, launch files, and robot descriptions without running ROS 2.

Category
Visit Server

README

ROS 2 Workspace Inspector MCP

License: Apache-2.0 Python 3.10+ CI

English | 简体中文

ROS 2 Workspace Inspector MCP is a local-first, read-only static-analysis MCP server that gives Codex, Claude Code, Cursor, and other MCP clients structured information about ROS 2 workspaces. Seven static-analysis tools cover discovery, package structure, dependencies, interfaces, launch files, robot descriptions, and bounded workspace diagnosis without requiring ROS 2. It also provides two compact read-only resources and one guided review prompt.

7 analysis tools · 2 read-only resource forms · 1 guided review prompt

The project inspects files only. It does not control robots, run ROS nodes, build workspaces, execute launch or Xacro files, execute setup.py, publish topics, call services or actions, or modify user files. All access stays inside the configured root sandbox. Only stdio is supported.

Tools

  • scan_workspace — Discover ROS 2 packages, workspace layout, build types, and basic manifest metadata.
  • inspect_package — Inspect a package's manifest, build configuration, executables, launch files, interfaces, robot descriptions, and tests.
  • analyze_dependencies — Analyze package dependencies, internal dependency edges, topological order, and dependency cycles.
  • inspect_interfaces — Parse and inspect ROS 2 Msg, Srv, and Action interface definitions.
  • analyze_launch_file — Statically analyze a Python, XML, or YAML ROS 2 launch file without executing it.
  • inspect_robot_description — Inspect URDF structure and statically summarize Xacro files without expanding or executing them.
  • diagnose_workspace — Run a consolidated, bounded diagnosis of the configured ROS 2 workspace.

Requirements and installation

  • Python 3.10 or newer
  • uv

Clone the repository, enter it, and create the locked environment:

git clone https://github.com/doer0312/ros2-workspace-mcp.git
cd ros2-workspace-mcp
uv sync --locked

Run against a local workspace:

uv run ros2-workspace-mcp --root /absolute/path/to/your/ros2_ws

The equivalent module entry point is:

uv run python -m ros2_workspace_mcp --root /absolute/path/to/your/ros2_ws

The process communicates via stdout using the MCP protocol. Do not redirect ordinary logs to stdout. Invalid CLI input is reported on stderr with a non-zero exit status.

Client configuration

Codex TOML:

[mcp_servers.ros2-workspace-inspector]
command = "uv"
args = ["--directory", "/absolute/path/to/ros2-workspace-mcp", "run", "ros2-workspace-mcp", "--root", "/absolute/path/to/your/ros2_ws"]

Claude Desktop JSON:

{
  "mcpServers": {
    "ros2-workspace-inspector": {
      "command": "uv",
      "args": ["--directory", "/absolute/path/to/ros2-workspace-mcp", "run", "ros2-workspace-mcp", "--root", "/absolute/path/to/your/ros2_ws"]
    }
  }
}

Copyable versions are available under examples/client-configs/. Both examples run the checked-out repository environment and do not assume a PyPI release.

Try the included static demo without installing ROS:

uv run ros2-workspace-mcp --root ./examples/demo_ws

Minimal request:

Scan the configured ROS 2 workspace and summarize its packages.

Guided review example:

Use review_ros2_workspace with depth="standard".

Available tools

The server exposes exactly seven tools:

  1. scan_workspace — discover packages and basic metadata.
  2. inspect_package — inspect one package's manifest, build files, executables, and file layout.
  3. analyze_dependencies — parse declarations and build a workspace dependency graph.
  4. inspect_interfaces — parse one package's .msg, .srv, and .action files.
  5. analyze_launch_file — statically analyze one Python, XML, or YAML launch file.
  6. inspect_robot_description — validate URDF or summarize unexpanded Xacro.
  7. diagnose_workspace — run a bounded, deduplicated workspace-wide diagnosis.

Example requests:

Use scan_workspace to inspect the configured ROS 2 workspace.
Inspect the package named demo_robot.
Analyze src/demo_robot/launch/demo.launch.py without running it.
Diagnose the configured workspace.

scan_workspace takes no arguments and scans only the workspace supplied with --root. A compact result looks like this:

{
  "layout": "colcon_workspace",
  "package_count": 1,
  "packages": [
    {
      "name": "demo_pkg",
      "relative_path": "src/demo_pkg",
      "build_type": "ament_python",
      "build_type_source": "inferred",
      "valid": true
    }
  ],
  "issues": []
}

The full result also includes server identity, canonical root and scan paths, package metadata, valid/invalid counts, duplicate names, and structured issues. See docs/tools.md.

Resources and prompt

  • ros2-workspace://summary is a fixed compact workspace summary.
  • ros2-workspace://package/{package_name} is a package context template with strict name validation and ambiguity errors.
  • review_ros2_workspace(focus="", depth="standard") guides quick, standard, or deep review using the seven tools without reading the workspace while the prompt is generated.

Resources are generated on read, have application/json MIME type, and never return raw project files. The server intentionally does not expose an arbitrary file resource, subscriptions, notifications, or filesystem watching. Client applications may display resources and prompts in different ways. See docs/resources.md and docs/prompts.md.

Security boundary

The server is bound to one validated, canonical workspace directory. Every file selected for reading is resolved and checked with path containment rather than string prefixes. Directory symlinks are never traversed; readable file symlinks must resolve inside the root. Text reads are UTF-8, size-limited, and centralized. Python is parsed with AST, XML with ElementTree, and YAML with safe_load; setup, launch, Xacro, CMake, and process declarations are never executed. See docs/security.md for the complete boundary.

Development

uv run ruff check .
uv run ruff format --check .
uv run pytest

Further documentation:

License

Licensed under the Apache License 2.0. See LICENSE.

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