ROS2 Control Execution

The skill provides a small setup script and a strict wrapper around 'ros2 control' subcommands. The requested actions (sourcing ROS, writing a local config.json, reading parameter profiles) match the stated purpose of executing and parameterizing ROS 2 control commands. There are no extraneous credentials, binaries, or unrelated config paths declared.

SKILL.md limits action to configuring and invoking the provided safe wrapper and explicitly requires sourcing the ROS environment. The wrapper enforces an allowlist of subcommands and uses subprocess without shell=True to avoid command-injection. Points to note: (1) it reads parameter/profile YAML files from ~/.openclaw/workspace/ros_profiles and will fail if those files are missing — these files are reasonable for profiles but are reads of user home data; (2) it sources the ROS setup.bash to construct the environment, which executes whatever that setup script contains (this is necessary for ROS but means you should trust your ROS installation).

This is an instruction-only skill with two small scripts included; there is no install spec that downloads or executes external code. Nothing is fetched from the network and no archive extraction is performed.

The skill does not request environment variables or credentials. It does require that the user has a functioning ROS 2 installation (ros2 on PATH and AMENT_PREFIX_PATH set) and writes a local config.json under the skill's directory. It also reads parameter/profile YAMLs from ~/.openclaw/workspace/ros_profiles when --profile is used. These accesses are proportionate to the feature but you should verify that profile files and your ROS setup are trusted.

always is false and the skill does not request persistent elevated privileges or attempt to modify other skills or global agent settings. It writes a small config.json under its own config directory (normal).