retopo
Skill2 starsChanged 26 days agoMITView on GitHub
Use when a high-poly or scanned mesh needs clean quad-dominant topology for texturing, rigging or rendering. Runs the surfacing-artist in a fresh context, retopologises headless in Blender with Shrinkwrap to the source, and reports face counts and deviation in mm.
In this repository30 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- agent-reviewSkill
- ai-mesh-cleanSkill
- cad-buildSkill
- couponSkill
- curriculumSkill
- dfm-reviewSkill
- fileSkill
- gddSkill
- ip-checkSkill
- led-budgetSkill
- lesson-learnedSkill
- makerworld-publishSkill
- material-pickSkill
- mcp-new-toolSkill
- ortho-sheetSkill
- pi-power-budgetSkill
- preprint-checkSkill
- promo-planSkill
- qa-gateSkill
- render-criticSkill
- research-oneSkill
- research-topicSkill
- safety-sheetSkill
- sculptSkill
- slicer-questionSkill
- style-bibleSkill
- tech-packSkill
- unit-costSkill
- verifySkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 19
Before stats with .venv\Scripts\python.exe and trimesh: faces, tri/quad/ngon counts, components, is_watertight, bbox in mm.
- Very specific: line 20
Write retopo.py for <blender_exe> --background --python retopo.py -- <hi path> <target faces> <out dir>.
- Distinctive: line 22
After stats, same columns as step 1, plus quad %, non-manifold edge count, UV status (none, or unwrapped with seams noted).
- Very specific: line 24
Write retopo.md; return paths, the before/after table, deviation numbers and anything the target could not meet.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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.
Files
Overview
Retopologise: $ARGUMENTS
Output on disk, in the project folder you name (see templates\project\), or work\retopo\<model-stem>\ if none is given: <model-stem>_lo.obj (plus .blend), retopo.py beside it, and retopo.md with the before/after table and deviation.
Blender comes from config\paths.yaml (blender_exe) through tools\mm_config.py. If it is empty, stop and say Blender is not configured.
When to use
The mesh will be textured, rigged or rendered and its topology is noisy (sculpt, scan, AI output). Not when the goal is a print: watertightness and minimum wall matter there, not quads; use /ai-mesh-clean. Not for a reference sheet (/ortho-sheet).
Steps
- Before stats with
.venv\Scripts\python.exeandtrimesh: faces, tri/quad/ngon counts, components,is_watertight, bbox in mm. If the "use" field says print, stop and return "use /ai-mesh-clean". - Write
retopo.pyfor<blender_exe> --background --python retopo.py -- <hi path> <target faces> <out dir>. Method per https://docs.blender.org/manual/en/latest/modeling/meshes/retopology.html: build the low-poly (remesh to a quad-dominant base, then reduce to the target), Shrinkwrap it to the high-poly, apply, recalculate normals. Record every parameter the script uses inretopo.md. - Deviation: sample the low-poly vertices against the high-poly surface with
trimesh.proximity.closest_point; report max and mean distance in mm and the percentage of samples above 0.5 % of the bbox diagonal. - After stats, same columns as step 1, plus quad %, non-manifold edge count, UV status (none, or unwrapped with seams noted).
- Render one shaded view of lo and hi side by side (headless; script beside the PNG) and open the PNG.
- Write
retopo.md; return paths, the before/after table, deviation numbers and anything the target could not meet.
Common rationalizations
| Rationalization | Reality |
| "Decimate to the target count and ship it." | Decimate leaves triangles and no loops. Report face-type counts; the target is quad-dominant topology, not a face count. |
| "Shrinkwrap applied, so it matches the source." | Measure it. Max and mean deviation in mm, or the report is an opinion. |
| "I'll tweak it in the Blender GUI." | GUI edits leave no record; retopo.py is the record. A model whose source script was lost cannot be rebuilt. |
| "The comparison render ran, no need to look." | Open the PNG. A preview you did not look at has misled builds before. |
Red flags (stop and fix)
- A report with a face count but no tri/quad/ngon split.
- Deviation missing or given without unit.
- Output written into a folder listed in
protected_roots(the guard hook blocks it). - A PNG never opened.
Verification
retopo.mdhas before/after rows: faces by type, components, watertight, bbox mm, quad %, non-manifold edges.- Max and mean deviation in mm reported from
trimesh.proximity.closest_point. retopo.pybeside the outputs and run this session with the parameters listed.- Comparison PNG exists, was opened, and is described in one line.
- Paths returned.
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
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.

