executing-plans
executing-plans
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available
ros2-operationsRoute ROS 2 CLI-backed graph inspection, bounded topic/service/parameter actions, doctor checks, and log inspection through ROSA.
Independent GitHub snapshot · Automated classification · Original documentation governs
Use this route when the task concerns ROS 2 (including Humble, Iron, or Jazzy),
ros2 node, ros2 topic, ros2 service, ros2 param, ros2 doctor, DDS or
domain discovery, or ROS 2 command and daemon errors.
This sub-skill describes ROSA's ROS 2 tools, not a general-purpose shell interface. Read the focused contracts before choosing a tool:
ROSA(ros_version=2, ...), model/tool-calling
setup, or general agent behavior, use the parent agent-core.rosnode or rostopic, use the sibling
ros1-operations. Do not translate a ROS 1
command into this route by guesswork.ros2_doctor() and the relevant list tool to observe the current graph;
never invent node, topic, service, or parameter names.info/echo/call, get a parameter before setting it, and
inspect a service type before calling it. Run tool calls one at a time and
wait for each result.count 1–10) and treat service calls and parameter
writes as potentially state-changing operations requiring an explicit,
validated target and request.subprocess.check_output(..., shell=True). It must not be used for arbitrary
command execution: the shallow category check does not block shell syntax
after an accepted prefix. Validate names, options, type tokens, and request
YAML; reject shell metacharacters and never suggest an arbitrary launcher.rclpy nor the ros2 executable, so this
route does not claim live ROS 2 verification.