Interface

ROS 2

The Robot Operating System 2, where agents write nodes, launch files and tests, or drive robots through topics and services.

Toolsfor ROS 2

The most active tools first.

  1. 2 mo ago

    ToolBB-0182github.com263 starsPython

    quackd: a Python CLI that drives registered robots with LLM-chosen skills

    quackd registers robots by name and has an LLM work toward a typed goal by calling one named skill per turn. Goals come from a terminal or from Claude over MCP, within step and time budgets, confirm gates and dry runs. The README lists eight robot bodies. Only a LeRobot SO-101 arm has run on hardware.

    Claimed
  2. 5 mo ago

    ToolBB-0045github.com3 starsPython

    TopicForge: a read-only MCP server for inspecting ROS 2 and DDS

    TopicForge gives an agent 15 read-only MCP tools for a ROS 2 graph, rosbag files and DDS discovery traffic. The code has no write path to the robot. In the README's example it finds a best-effort publisher facing a reliable subscriber. CI runs it against a ROS 2 Humble and Jazzy test bench; no physical robot is shown.

Builds and news tagged ROS 2

Ranked by points weighed against age. New entries with early votes rise.

  1. 21 d ago

    BuildBB-0233github.com

    ETH Zurich project has a language model set rehab robot parameters over MCP

    Celal Türkmen's ETH Zurich semester project has a language model convert a therapist's plain-language request into an MCP tool call for a rehabilitation robot. A validator built from each device's YAML file of parameters, units and bounds decides whether the call reaches the hardware. The README reports runs on a Novint Falcon haptic device under ROS 2 Jazzy.

  2. 3 mo ago

    NewsBB-0126arxiv.org

    Paper: coding agents write robot behavior trees from an MCP skill contract and run them on a Husarion Panther

    Researchers had coding agents write robot behavior trees from a contract served by a robot-side MCP server. The contract lists the robot's skills, allowed operators and optional templates, and a gate validates each tree before it runs. The paper reports tests of Sonnet 4.6 and Gemma4:31b on 110 simulated tasks and 14 on a Husarion Panther running Nav2.

    Claimed