Skip to content
Back to skills

Ros2 Operations

ASecurity

"Route ROS 2 CLI-backed graph inspection, bounded

  • 247 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 9, 2026
toolsrustgoshellnodeapi

Works with

  • cli
  • api

Security analysis

A100/100

Pro scans all 4 files and shows the line behind each finding

Scanned September 9, 2026

npx -y skills add VectorSpaceLab/AREX-Skill --skill ros2-operations --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Ros2 Operations?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Ros2 Operations
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vectorspacelab-ros2-operations/badge)](https://www.skillsdirectory.com/skills/vectorspacelab-ros2-operations)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ros2-operations
description: "Route ROS 2 CLI-backed graph inspection, bounded
  topic/service/parameter actions, doctor checks, and log inspection through
  ROSA."
disable-model-invocation: true
metadata:
  disco-role: operating
license: Apache 2.0
---

# ROS 2 operations

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:

- [API reference](references/api-reference.md) — arguments, command shapes,
  filtering, return values, and failure values.
- [Workflows](references/workflows.md) — source the intended distro, discover
  before acting, and keep operations bounded and sequential.
- [Troubleshooting](references/troubleshooting.md) — dependency, discovery,
  quoting, timeout, parser, and log failures.

## Route boundaries

- For constructing or invoking `ROSA(ros_version=2, ...)`, model/tool-calling
  setup, or general agent behavior, use the parent [agent-core](../agent-core/SKILL.md).
- For ROS 1/Noetic APIs such as `rosnode` or `rostopic`, use the sibling
  [ros1-operations](../ros1-operations/SKILL.md). Do not translate a ROS 1
  command into this route by guesswork.
- For blacklists, deterministic math, custom LangChain tools, or robot-specific
  prompts, use [tool-customization](../tool-customization/SKILL.md).
- For shared installation and runtime prerequisites, start at the parent
  [ROSA route](../../SKILL.md).

## Operating rules

1. Establish which ROS 2 distro and shell environment are intended. ROSA does
   not source a distro for the caller.
2. Use `ros2_doctor()` and the relevant list tool to observe the current graph;
   never invent node, topic, service, or parameter names.
3. List before `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.
4. Keep topic echo bounded (`count` 1–10) and treat service calls and parameter
   writes as potentially state-changing operations requiring an explicit,
   validated target and request.
5. Treat every value interpolated into a ROS 2 command as untrusted input. The
   helper checks only the first command categories, then invokes
   `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.
6. Distinguish mocked wrapper tests from live middleware observations. The
   construction host has neither `rclpy` nor the `ros2` executable, so this
   route does not claim live ROS 2 verification.

Files in this skill

  • SKILL.md2.9 KB
  • references/api-reference.md11.4 KB
  • references/troubleshooting.md9.4 KB
  • references/workflows.md8 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.