openamrobot-harness

Plugin openamrobot-harness from marketplace openamrobot in openAMRobot/.github. Commands read from the repo's manifest — 5 days ago.

Skills (16)

• harness-ci-hardening

  How to make every OpenAMRobot repository build and run all its tests in GitHub Actions reproducibly and quickly, using the organization's reusable workflow. Use for CI tasks, for issues like platform-sw #98, and when a repository is prepared for build missions.

• harness-dependency-hygiene

  How to audit and fix declared versus used dependencies, pins, vulnerabilities and licences in OpenAMRobot ROS 2, Python, Node and firmware repositories. Use for the weekly dependencies audit and whenever a PR changes package.xml, requirements, package.json, .repos or CMakeLists.txt.

• harness-embodied-ai-pipeline

  How harness agents work on the OpenAMRobot demonstration-learning stack (teleoperation recording, datasets, policy training, evaluation, policy runner). Use for any task in openamrobot-embodied-ai, the data pipeline or policy evaluation.

• harness-evidence-and-pr

  How an OpenAMRobot harness agent proves its work and writes the PR body. Use for every build mission and every review, before writing .harness-out/pr.md or a review.

• harness-firmware-review

  Review checklist and host-test approach for the OpenAMRobot STM32H723 base controller and upper-body firmware (micro-ROS over UDP). Agents review and write host tests only; firmware is tier C. Use when reviewing a firmware PR or auditing firmware repositories.

• harness-impact-analysis

  How to find what a change affects across packages, repositories, topics and register values, and which tests must run. Use in every review, in the impact audit, and before a build mission finishes.

• harness-maintainability

  How to measure and improve maintainability (complexity, duplication, dead code, test coverage, documentation drift) in OpenAMRobot repositories with small, safe refactors. Use for the weekly maintainability audit and refactor tasks.

• harness-performance

  How to measure and improve runtime performance of OpenAMRobot ROS 2 nodes on the Jetson Orin NX target, using simulation and CI as proxies, with regression rules. Use for the weekly performance audit and for optimisation tasks.

• harness-protected-paths

  What a harness agent must never change on a mobile manipulator, how to test safety-relevant code without changing it, and when to stop and hand over. Use whenever a task touches motion, stopping, power, arms, the lift, firmware, contracts or any tier C or D path.

• harness-register-and-baseline

  How harness agents use the decisions register and the plan-set baseline as ground truth for every technical value. Use whenever a task involves a dimension, frame, sensor, part, limit, topic name, date or any value that could be in decisions.yaml.

• harness-robot-description-fidelity

  How to keep the OpenAMRobot Xacro/URDF, ros2_control and simulation models traceable to the register, CAD, supplier data and measurements. Use when touching robot descriptions, sensor frames, joint limits, inertias, ros2_control tags or Gazebo models, or when reviewing such a change.

• harness-ros2-code-quality

  Review and improvement checklist for production-grade ROS 2 Jazzy nodes on a mobile manipulator (lifecycle, QoS, executors, parameters, time, frames, errors, logging). Use when writing, refactoring or reviewing node code, launch files or configuration.

• harness-ros2-contracts

  How to check message, service, action, topic, QoS, parameter and launch-argument consistency across all OpenAMRobot repositories. Use for the contracts audit, for reviewing any PR that publishes, subscribes or calls an interface, and before proposing any interface change.

• harness-ros2-testing

  How to write and run tests for ROS 2 Jazzy packages in OpenAMRobot (pytest, gtest, launch_testing, colcon), including flake control and the test pyramid. Use whenever a mission adds, fixes or runs tests in a ROS 2 repository.

• harness-simulation-validation

  How to write, run and judge headless Gazebo simulation scenarios for the OpenAMRobot base and upper body in CI, and what simulation results may and may not claim. Use when adding scenarios, analysing nightly simulation results or reviewing a claim that cites simulation.

• harness-static-analysis

  Static analysis and lint tools for OpenAMRobot code (Python, C++, launch, YAML, workflows, shell, firmware C) and how to introduce them without breaking CI. Use for lint-fix tasks, CI hardening and review.

Hooks (1)

• SessionStart

  cat "${CLAUDE_PLUGIN_ROOT}/CORE_RULES.md"