estampo

name	estampo
description	Reproducible 3D print pipelines with estampo. Use when asked to print, slice, or set up printing for a project's 3D models (STL/STEP/3MF or code-CAD output) — creating or editing estampo.toml, choosing slicer engine and profiles, tuning overrides for supports/adhesion/thin walls, running the pipeline, and pinning everything for CI.

Print with estampo

Use this skill when asked to print or slice parts, automate printing, or set up a print pipeline in a project that has 3D models (STL, STEP, 3MF) or code-CAD sources that produce them. estampo is a declarative build system: everything — parts, slicer settings, plate layout, post-processing — lives in one estampo.toml, so the whole job is reviewable text, not GUI clicks.

Reference material when you need detail beyond this skill:

• llm.md — config reference with the common override keys for both engines
• estampo.schema.json — full config schema; orca-settings.json / cura-settings.json for every setting

---

Step 0 — Check the environment

estampo --version || uv tool install estampo   # or: pip install estampo
docker info --format 'docker ok' 2>/dev/null

Docker is how estampo pins the slicer version for identical G-code everywhere; without it a local slicer install is used as a fallback (not reproducible — warn the user). If the project already has an estampo.toml, read it first and edit rather than regenerate.

Step 1 — Find the parts

Locate STL/STEP/3MF files, and the code-CAD sources that generate them (build123d/CadQuery Python, OpenSCAD). If meshes are generated, regenerate them from source before slicing rather than trusting stale artifacts. STEP is preferred input from code-CAD — estampo tessellates it at load time.

Step 2 — Engine and printer (ask, don't guess)

The one rule: orca for Bambu Lab printers, cura for everything else. The bundled Orca machine profiles are all Bambu Lab; the bundled Cura set covers hundreds of other printers (Creality, Prusa, Voron, Ultimaker, Anycubic, Elegoo, …). Picking orca for a non-BBL printer means the user must supply their own machine profile.

[ASK: Which printer (make/model/nozzle) and which filament(s) are loaded?] if the repo doesn't say. Then discover exact profile names — never invent them:

estampo profiles list --engine orca --category machine
estampo profiles list --engine orca --category process --printer "Bambu Lab P1S 0.4 nozzle"
estampo profiles list --engine orca --category filament --printer "Bambu Lab P1S 0.4 nozzle"

With orca, always pass --printer when listing process/filament profiles (the filter is OrcaSlicer-only — cura ignores it). OrcaSlicer silently fails at slice time (exit 239) if the process and printer are incompatible; the filter only shows compatible ones.

Filament rule: match the profile to the filament actually loaded, not the printer brand. Vendor profiles (e.g. "Bambu PLA Basic") assume that vendor's filament and under-extrude with generic spools — default to "Generic … @base".

Step 3 — Write and validate the config

name = "widget"

[slicer]
engine = "orca"
version = "2.3.1"              # pin for reproducibility

[slicer.orca]
printer = "Bambu Lab P1S 0.4 nozzle"
process = "0.20mm Standard @BBL X1C"
filaments = ["Generic PLA @base"]

[[parts]]
file = "widget.step"
orient = "upright"             # code-CAD output is usually Z-up already
copies = 1
filament = 1                   # AMS slot / extruder index

Per-part options: orient = "flat"|"upright", rotate = [x, y, z] degrees, copies, filament. Then:

estampo validate

validate checks structure and setting names — not print safety. Never tell the user the settings are safe; tell them to review the sliced output.

Step 4 — Overrides: map part concerns to settings

Put process tweaks in [slicer.orca.overrides] (or [slicer.cura.overrides] — key names differ between engines; see llm.md for both tables). The common mappings from part geometry to settings:

Concern	Orca override
Overhangs > ~45°	enable_support = 1, support_type = "tree(auto)", support_threshold_angle = 45
Poor bed adhesion / tip-over risk	brim_type = "outer_only", brim_width = "5mm"
Thin walls near nozzle width	detect_thin_wall = 1, check line_width
Strength	wall_loops = 3, sparse_infill_density = "25%"
Dimensional accuracy (holes tight)	xy_hole_compensation, elefant_foot_compensation

If the geometry came from the build123d-mcp server, its analyze_printability() report (overhangs, thin walls, brim/raft need, bed-fit) tells you exactly which of these to set — apply the findings here instead of guessing.

Step 5 — Run

estampo run

Read the summary it prints (part count, plate fit, print time, filament use) back to the user. Useful variants: estampo run --until slice (stop before post-processing), estampo run --only arrange (re-run one stage). Output lands in estampo_output/ — keep it gitignored.

Step 6 — Make it reproducible (and CI, if asked)

1. version = "…" pinned in [slicer] — exact slicer binary via Docker.
2. estampo profiles pin — copies referenced profiles into ./profiles/; commit that directory so builds don't depend on locally installed profiles.
3. Commit estampo.toml + CAD sources + (optionally) generated meshes; gitignore estampo_output/.

For CI, estampo init --workflow scaffolds a GitHub Actions workflow, or write one that regenerates meshes from code-CAD sources first and then runs estampo run — same repo, same config, same G-code.

---

Pitfalls

• Exit 239 from OrcaSlicer = incompatible printer/process/filament profile combination. Re-run estampo profiles list … --printer "<printer>" and pick from the filtered list.
• Vendor-branded filament profiles with generic filament cause under-extrusion — use "Generic … @base" unless the actual spool matches.
• Slicing without Docker works but is not reproducible across machines; say so rather than silently falling back.
• estampo generates G-code from configuration but cannot verify it is safe for the user's specific printer — always tell the user to review the sliced output before printing.