Authoring and Validating URDF Robot Descriptions

Where this fits

You just asked your agent to generate a URDF for a six-axis arm. The links look right, but when you load it into RViz, the second joint rotates around the wrong axis. You start tracing frames and realize the authoring step skipped validation. This skill gives an agent a concrete checklist for creating, inspecting, and debugging .urdf robot descriptions—so kinematic mistakes are caught before simulation, not after.

Why agents benefit

• The skill treats URDF work as constrained kinematic modeling, not generic XML, so agents check frame placement, joint-axis semantics, units, mesh scale, and inertial data.
• It instructs agents to author and edit URDF XML directly, avoiding a brittle Python generation pipeline and a gen_urdf() contract.
• It scopes correctly: use the SRDF skill for MoveIt2 groups and IK semantics, the cad-viewer skill for local server controls, and the CAD skill for geometry formats.
• Local installed skill files are the runtime source of truth, with the GitHub repo used only for provenance, making agent behavior reproducible.
• It surfaces common failure modes: invalid joint limits, missing collision geometry, mismatched reference frames, and broken mesh references.

Practical scenarios

Inspecting and repairing a misaligned joint

A developer spots a robot arm where the shoulder joint spins around Z instead of the expected Y. The agent reads the .urdf, checks the joint axis and origin relative to parent and child link frames, then writes a corrected XML snippet with exact coordinates.

Validating inertial and mesh scale

A URDF loads in Gazebo but the base link vibrates and floats. The agent inspects the inertial tensor and mesh references, confirms the STL/GLB units match the link frame, and flags missing or invalid collision geometry.

Building a robot from scratch

A user asks for a two-finger gripper. The agent creates a new .urdf directly, defines links, joints, their limits, visual/collision geometry, and then runs the skill's validation rules before presenting the file.

Add it to your agent workflow

Install the skill with:

npx skills add earthtojake/text-to-cad --skill urdf

Then prompt: "Create a URDF for a 6-DOF robot arm and validate frame conventions, joint limits, and inertial data." The agent will use the installed local skill files as the runtime source of truth. For MoveIt2 semantic groups, pair this skill with SRDF; for STEP/STL/3MF outputs, pair it with CAD. You can also browse the skill page at OpenAgentSkill: https://www.openagentskill.com/skills/earthtojake-urdf

Compare before adopting

Before adding this skill to your stack, compare it with generic XML authoring skills. Look at whether it provides concrete quality signals such as mesh-reference checking and inertial plausibility. Check maintenance freshness relative to URDF changes. Decide if it fits your workflow—if you already have strong validation tooling, this may overlap. For most agent use, a focused URDF skill beats a general XML skill for frame errors.

Why it is worth tracking

The skill is maintained in earthtojake/text-to-cad, a 14k-star repository, and it is tagged in the research category with a narrow, useful scope. The strongest quality signal is that it encodes domain-specific validation rules that general XML agents miss. Track it if you regularly write robot descriptions; re-evaluate whenever the companion CAD/SRDF skills or local skill files change. A targeted, actively maintained skill like this one is more valuable than an overgrown prompt in a toolchain where correctness depends on frames and units.