CtrlK
BlogDocsLog inGet started
Tessl Logo

cad-workbench

Use for persistent, visual, short-loop CAD modeling with cad.open, cad.add_part, cad.edit_part, cad.place, cad.import, cad.delete, cad.look, cad.inspect, Host-owned CAD Review, and cad.export.

67

Quality

84%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

88%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A strong operational skill body: concrete dispatch rules, executable Build123d examples, and a tightly validated edit-inspect-repair loop with explicit error-recovery paths. Its only weaknesses are repeated full tool-call syntax that inflates tokens and some long inline policy sections that could move to a reference file.

DimensionReasoningScore

Conciseness

The body is dense and operational with no padding or explanations of known concepts — every line carries a build contract or decision rule. It falls short of a 5 only because full call syntax like "cad.edit` `action=\"add_part\"" is repeated verbatim many times (Loop, Python Cell Contract, Deleting sections) where a once-defined shorthand would trim tokens.

4 / 5

Actionability

Guidance is fully executable: concrete dispatch rules ("The part is not in the scene → cad.edit action=\"add_part\""), exact semantics ("pass the position you want, not a delta", "sets an object's absolute position and rotation"), and copy-paste-ready code ("body = Box(40, 30, 12)", "publish(\"body\", body, \"Main body\")", "body_width = param(\"body_width\", 40.0)") covering the common cases.

5 / 5

Workflow Clarity

The Loop section sequences six steps with explicit validation checkpoints ("Inspect the attached preview directly before doing anything else", "if the preview shows anything else, stop and read the scene with cad.inspect") and complete feedback loops for error recovery (previewError handling, stale_build recovery, focused repair). Destructive deletes are also gated: "Deletion commits a normal build and requires a handoff cad.export" with the Undo fallback, so the destructive-operation cap does not apply.

5 / 5

Progressive Disclosure

Sections are well organized (Scene, Placement, Loop, Python Cell Contract, Deleting, Project PCB inputs, Observation Rules, Build Discipline) with one clearly signaled external reference ("Load build123d-core only when specific API syntax or modeling details are needed") and no nested references. Minor gap: longer policy material such as the PCB-comparison checklist is inlined in a ~156-line file where a one-level reference file could carry the detail.

4 / 5

Total

18

/

20

Passed

Description

71%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A specific, distinctive description that names its domain and a full tool inventory, but it reads as a tool manifest rather than a capability statement: it lacks natural trigger synonyms (3D model, part, assembly, STEP) and concrete when-to-use phrases a user would actually say.

Suggestions

Add plain-language capability verbs to the what-clause, e.g. "Create and edit parametric 3D parts, assemble and place them in one persistent scene, and export STEP files for review".

Broaden trigger terms beyond "CAD modeling" with natural synonyms and file formats users say: "3D model", "part design", "assembly", "enclosure", "STEP files".

Make the when-clause concrete with user-shaped triggers, e.g. "Use when the user asks to model, modify, or review a mechanical part, enclosure, or assembly".

DimensionReasoningScore

Specificity

The description names a concrete domain ("persistent, visual, short-loop CAD modeling") and enumerates ten specific tools ("cad.open, cad.add_part, cad.edit_part, cad.place, cad.import, cad.delete, cad.look, cad.inspect, ... cad.export"), giving several specific actions with minor gaps in plain-language capability verbs like "export STEP files" or "place components".

4 / 5

Completeness

Both parts are present: "Use for ... CAD modeling" is an explicit when-clause and the tool enumeration states the what. However, the when lacks concrete user-trigger phrases (e.g., "when the user asks to model a part or design an enclosure") and the what lists tools rather than outcomes, so it falls short of the anchor-5 example.

4 / 5

Trigger Term Quality

"CAD modeling" is the natural core trigger a user would say, but common variations and synonyms are missing ("3D model", "part design", "assembly", "enclosure", "STEP"), and the rest of the description is tool-name jargon ("cad.open", "Host-owned CAD Review") users would not naturally say.

3 / 5

Distinctiveness Conflict Risk

The description occupies a clear niche — a host CAD workbench with proprietary cad.* tool names and "Host-owned CAD Review" — so it is clearly distinguishable from generic document, file, or data skills with minimal conflict risk.

5 / 5

Total

16

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
JimmyPang02/open-replicator
Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.