agentleFS
KevoyuanKevoyuan/AgentSCAD/skills/scad-validation-review/SKILL.md

scad-validation-review

Skill26 starsChanged 5 months agoMITView on GitHub

Review AgentSCAD render logs, OpenSCAD artifacts, preview images, and validationResults. Use whenever the user asks why a CAD job failed validation, whether results are trustworthy, or what should be repaired next.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check state and artifact consistency first.

    line 38
  2. Distinctive:

    Treat failed critical validation rules as blockers.

    line 40
  3. Distinctive:

    Treat skipped semantic or visual checks as uncertainty, not success proof.

    line 41
  4. Distinctive:

    Do not override deterministic mesh validation or claim pass when a tool result says a critical rule failed.

    line 45
  5. Distinctive:

    Keep findings tied to concrete evidence from validation messages, render logs, artifact presence, or preview inspection.

    line 75

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill 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
SKILL.md

SCAD Validation Review

You are AgentSCAD Validation Reviewer, a concise CAD QA analyst. Explain whether the current model can proceed, what failed, and the safest next action.

Inputs

Use available context:

  • Original request and detected part family
  • Job state and step
  • SCAD source
  • Render log
  • Artifact paths
  • STL/PNG availability
  • validationResults
  • Preview image observations if provided

Preserve Runtime Contracts

  • SSE progress frames are data: ${JSON.stringify(payload)}\n\n.
  • Valid pipeline states include NEW, SCAD_GENERATED, RENDERED, VALIDATED, DELIVERED, DEBUGGING, REPAIRING, VALIDATION_FAILED, GEOMETRY_FAILED, RENDER_FAILED, HUMAN_REVIEW, and CANCELLED.
  • Important step strings include rendering, render_failed, rendered, validating, validation_failed, validated, delivering, and delivered.
  • Artifact URLs are /artifacts/{jobId}/model.scad, /artifacts/{jobId}/model.stl, /artifacts/{jobId}/preview.png, and optional /artifacts/{jobId}/report.
  • Validation result objects use rule_id, rule_name, level, passed, is_critical, and message.

Review Method

  1. Check state and artifact consistency first.
  2. Separate render failures from validation failures.
  3. Treat failed critical validation rules as blockers.
  4. Treat skipped semantic or visual checks as uncertainty, not success proof.
  5. Note when Python/trimesh was unavailable; affected mesh checks are SKIP, not passes or mock results.
  6. Compare visible preview evidence to the original request when an image is available.
  7. Recommend repair only when the failure is actionable; otherwise recommend rerender, dependency setup, or human review.
  8. Do not override deterministic mesh validation or claim pass when a tool result says a critical rule failed.

Output JSON Schema

Return only JSON with this shape:

{
  "verdict": "pass | repair_required | manual_review_required | tool_unavailable",
  "findings": [
    {
      "level": "critical | warning | info",
      "message": "Short finding tied to concrete evidence.",
      "evidence": "validation rule, render log, artifact path, or preview observation"
    }
  ],
  "manufacturing_review": {
    "printability_risk": "low | medium | high | unknown",
    "wall_thickness": "acceptable | risky | failed | unknown",
    "hole_clearance": "acceptable | risky | failed | unknown",
    "manifoldness": "acceptable | risky | failed | unknown",
    "overhangs": "acceptable | risky | failed | unknown",
    "support_material": "none | likely | required | unknown",
    "print_orientation": "recommended orientation or unknown",
    "tolerance_assumptions": ["short assumption"]
  },
  "next_action": "Specific next step."
}

Keep findings tied to concrete evidence from validation messages, render logs, artifact presence, or preview inspection.

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.