ROS 2 Skill

**ros2-skill 1.0.6** — Adds internal structure, enhanced documentation, and technical references. - Added extensive CLI documentation and workflow references (references/CLI.md, EXAMPLES.md, RULES.md). - Introduced internal Python modules for each major ROS 2 operation (e.g., `ros2_bag.py`, `ros2_component.py`, etc.). - Improved agent usage guidance and modular architecture via updated SKILL.md. - Transitioned from a single `EXAMPLES.md` file to a comprehensive, multi-file documentation system. - Expanded test coverage and internal caching through new pycache files. - Cleaned up obsolete files to align with new modular design.

**Summary:** Major update with extensive refactor, new agent decision rules, and improved safety and introspection for all ROS 2 robot control tasks. - Added strict, documented agent rules enforcing introspection before any robot action (no assumptions/hardcoded topic or message names). - Introduced mandatory session-start health, time, and node checks for robust operation. - Expanded and clarified usage to cover all ROS 2 operations: topics, services, actions, parameters, nodes, diagnostics, controllers, and safety checks. - Deprecated/removed previous "examples" and test files now replaced by built-in guided workflows and canonical command documentation. - Updated description and interface for comprehensive coverage and safer, more reliable ROS 2 interaction.

Initial release of ros2-skill — an adaptation of [ros-skill](https://github.com/lpigeon/ros-skill) redesigned for direct local ROS 2 communication via rclpy instead of rosbridge. ### Topics - `topics list` / `ls` — list all active topics - `topics type` — get the message type of a topic - `topics details` / `info` — publishers, subscribers, and QoS for a topic - `topics message` / `message-structure` / `message-struct` — introspect message field structure - `topics subscribe` / `echo` / `sub` — collect messages; `--duration` + `--max-messages` for batch collection - `topics publish` / `pub` / `publish-continuous` — single-shot or timed publish at `--rate` Hz - `topics publish-sequence` / `pub-seq` — publish a sequence of messages with per-step durations - `topics publish-until` — publish while monitoring a separate topic; stops when a condition is met (`--delta`, `--above`, `--below`, `--equals`); `--euclidean` for N-dimensional distance across multiple fields - `topics hz` — measure publish rate (rate, min/max/std_dev of inter-message intervals) - `topics bw` — measure topic bandwidth (bytes/s, bytes per message) - `topics delay` — measure end-to-end latency via `header.stamp` - `topics find` — find all topics publishing a given message type ### Nodes - `nodes list` / `ls` — list all active nodes - `nodes details` / `info` — publishers, subscribers, services, action servers, and action clients for a node ### Services - `services list` / `ls` — list all services - `services details` / `info` — request and response field structure for a service - `services call` — call a service with a JSON request - `services find` — find all services of a given type - `services echo` — echo service request/response event pairs (requires introspection enabled on the node) ### Parameters - `params list` / `ls` — list all parameters on a node - `params get` — get a parameter value - `params set` — set a parameter value - `params describe` — describe a parameter (type, constraints, read-only flag) - `params dump` — bulk-export all parameters for a node as JSON - `params load` — bulk-set parameters from a JSON string or file - `params delete` — delete one or more parameters ### Actions - `actions list` / `ls` — list all action servers - `actions details` / `info` — goal, result, and feedback structure for an action server - `actions type` — get the action type of an action server - `actions send` / `send-goal` — send a goal; `--feedback` streams feedback messages in the output - `actions find` — find all action servers of a given action type - `actions echo` — echo live feedback and status messages from an action server - `actions cancel` — cancel all in-flight goals on an action server ### Utilities - `version` — detect ROS 2 distro and domain ID - `estop` — emergency stop; auto-detects velocity topic and publishes zero velocity --- For the original ros-skill (ROS 1 + ROS 2 via rosbridge), see: [ros-skill](https://github.com/lpigeon/ros-skill)

- Internal script update in `scripts/ros2_cli.py`. - No changes to user-facing features or documentation.

1.0.1 Added: - Emergency stop command (`estop`) - auto-detects velocity topic and message type for mobile robots - Better error messages with hints and suggestions for unknown message types Changed: - Topics subscribe/publish/publish-sequence now auto-detect message type from topic if not provided - Updated documentation with new estop command Fixed: - Fixed infinite recursion bug in ROS2CLI class - Fixed misplaced imports and cleaned up duplicate imports - Fixed null pointer checks for publisher creation - Replaced bare except clauses with proper exception handling - Removed unused code and variables

Initial release: Fork of https://clawhub.ai/lpigeon/ros-skill, but updated to support only ROS 2, locally. No ROS 1, no rosbridge This release marks a major shift to ROS 2 only. The CLI has been renamed from ros_cli.py to ros2_cli.py, and rosbridge WebSocket communication has been replaced with direct rclpy integration for local ROS 2 control. ROS 1 support has been completely removed. Commands are simplified—no more --ip/--port flags needed since ROS 2 runs locally. Dependencies changed from websocket-client to rclpy and rosidl-runtime-py. All documentation has been updated to reflect the new architecture.