agentleFS
cedrickcheecedrickchee/text-to-cad/AGENTS.md

text-to-cad

AGENTS.md35 starsChanged 6 months agoMITView on GitHub

This repository is a harness for script-driven CAD generation with coding agents like Codex and Claude Code. If you are modifying CAD Explorer itself, go to .agents/skills/cad/explorer/README.md. Use the bundled skills for workflow details: Use the URDF skill when generating or editing a robot description. If an existing valid URDF already exists, the robot-motion skill can attach motion artifacts to it directly; use robot-motion for inverse kinematics, path planning, motion artifact generation, and motion-server testing. AGENTS.md is intentionally harness-focused. Reusable…

  • Installs packages
  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,439 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository4 other files it gives its agents.

The file

Show its rules and text

We split this AGENTS.md into the rules it gives an agent

13 rules. About 0 are common, 12 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If you are modifying CAD Explorer itself, go to .agents/skills/cad/explorer/README.md.

    line 5
  2. Distinctive:

    Use the bundled skills for workflow details:

    line 9
  3. Distinctive:

    Use the URDF skill when generating or editing a robot description.

    line 15
  4. Distinctive:

    Prefer the repo-local CAD runtime when it exists:

    line 39
  5. Distinctive:

    Do not hand-edit derived artifacts unless explicitly instructed.

    line 59
  6. Distinctive:

    If regenerated output differs from checked-in generated files, the regenerated output is authoritative.

    line 60
  7. Distinctive:

    Keep project CAD files in explicit repo-relative locations.

    line 64
  8. Distinctive:

    Use explicit generation targets.

    line 65
  9. Distinctive:

    Update project-local documentation only when project focus, entry roles, inventory, dependency notes, durable quirks, or preferred rebuild …

    line 67
  10. Distinctive:

    Never disable LFS filters for git add, commits, or other object-writing operations.

    line 70
  11. Distinctive:

    Start with the narrowest source-only search that can identify directly affected files.

    line 74
  12. Distinctive:

    If the first pass makes scope clear, edit the source first and validate after.

    line 76
  13. Distinctive:

    Do not run mutable generation, inspection, and render/review steps in parallel against geometry that is still changing in the same edit …

    line 77

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md

This repository is a harness for script-driven CAD generation with coding agents like Codex and Claude Code.

If you are modifying CAD Explorer itself, go to .agents/skills/cad/explorer/README.md.

Skill Routing

Use the bundled skills for workflow details:

  • .agents/skills/cad/SKILL.md for STEP, STL, 3MF, DXF, GLB/topology artifacts, render images, and @cad[...] prompt references.
  • .agents/skills/urdf/SKILL.md for generated URDF files, gen_urdf(), robot links, joints, limits, and URDF mesh references.
  • .agents/skills/robot-motion/SKILL.md for ROS 2/MoveIt dependency setup, running the motion server, and generating IK/path-planning artifacts for an existing URDF.

Use the URDF skill when generating or editing a robot description. If an existing valid URDF already exists, the robot-motion skill can attach motion artifacts to it directly; use robot-motion for inverse kinematics, path planning, motion artifact generation, and motion-server testing.

AGENTS.md is intentionally harness-focused. Reusable CAD, URDF, and robot-motion workflow rules live inside the skills.

Harness Context

Project CAD files are repo-relative. This harness does not reserve a project-file directory. Project CAD entries may live at the repository root under folders such as STEP/, STL/, DXF/, and 3MF/, or in another explicit repo-relative layout chosen by the project.

The CAD and URDF skill tools are file-targeted. They do not depend on a harness layout or prepend a project root.

Project-specific context may live in compact root-level notes such as PROJECT.md. Do not copy reusable generator contracts, prompt-reference rules, validation policy, Explorer/link rules, image review policy, or full CLI syntax into them; link to the relevant skill references instead.

For CAD Explorer scan-root, link, and @cad[...] behavior, defer to .agents/skills/cad/SKILL.md and its Explorer/rendering references.

Python Environment

Prefer the repo-local CAD runtime when it exists:

./.venv/bin/python

This environment has the CAD dependencies required by the skill tools, including build123d and OCP. If .venv is missing or cannot import those modules, create/install it from the repo root before running CAD tools:

python3.11 -m venv .venv
./.venv/bin/pip install -r .agents/skills/cad/requirements.txt

Other bundled skills own their Python dependencies in their skill directories; install them only when using those workflows.

Source Of Truth

  • Generated CAD, URDF, robot-motion outputs, Explorer sidecars, renders, topology, meshes, and flat-pattern artifacts are derived artifacts.
  • Do not hand-edit derived artifacts unless explicitly instructed. Edit the owning source file or imported source file first, then regenerate explicit targets with the relevant skill tool.
  • If regenerated output differs from checked-in generated files, the regenerated output is authoritative.

Repo Policies

  • Keep project CAD files in explicit repo-relative locations.
  • Use explicit generation targets. Do not run directory-wide generation.
  • Generation tools write and overwrite current configured outputs. They do not delete stale outputs when paths change.
  • Update project-local documentation only when project focus, entry roles, inventory, dependency notes, durable quirks, or preferred rebuild roots change.
  • CAD outputs are often LFS-tracked, and broad status checks can invoke LFS clean filters while generated files are changing; prefer path-limited git status during CAD work.
  • For bookkeeping-only full status, use git -c filter.lfs.clean= -c filter.lfs.smudge= -c filter.lfs.process= -c filter.lfs.required=false status --short.
  • Never disable LFS filters for git add, commits, or other object-writing operations.

Execution Notes

  • Start with the narrowest source-only search that can identify directly affected files.
  • Exclude generated artifacts, binary CAD files, caches, and build outputs from default searches unless the task explicitly targets them.
  • If the first pass makes scope clear, edit the source first and validate after.
  • Do not run mutable generation, inspection, and render/review steps in parallel against geometry that is still changing in the same edit loop. Rebuild first, then inspect, then review.
  • In cloud or constrained environments, avoid full-repo hydration when affected entries are known. Fetch only the needed inputs, generated outputs, and LFS objects for the entries being edited and explicitly regenerated.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.