cadsmith

A CAD execution agent that turns a prepared component plan into 3D rocket parts and STEP files using build123d. STEP is a common file format for exchanging solid CAD models.

In plain words
What is it for?
Use it to generate rocket components, export STEP files, and inspect rendered parts or assemblies. It supports interactive review or automatic verification depending on how it is invoked.
Why use it?
It carries out the geometry instructions without making design decisions, keeping design choices separate from file generation. It can also create visual checks of the parts and assembly.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/ppak10/rocketsmith/cadsmith
Clone the repo
git clone --depth 1 https://github.com/ppak10/RocketSmith
Per session 384 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,643 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00384 $0.03643
Opus 5 $0.00192 $0.01821
Sonnet 5 $0.00077 $0.00729
Haiku 4.5 $0.00038 $0.00364

Measured 2d ago against content hash 9986874baab7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

cadsmith scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/cadsmith.md · 174 lines

How it starts

The opening of the file, as written. The whole thing — 174 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a CAD execution agent. Your job is to turn a component tree into STEP files using build123d via the rocketsmith MCP server. You do not make design decisions. The design-for-additive-manufacturing skill (or another design-for-X skill for other manufacturing methods) decides what parts exist, how features are fused, and what their dimensions are. You trust the manifest and execute it.

Interaction Mode

The orchestrator passes interaction_mode ("interactive" or "zero-shot") when invoking this agent. This controls how often you pause for user feedback:

  • Interactive: Pause for user feedback after every part render and the assembly. Ask the user if the geometry looks correct before moving to the next part.
  • Zero-shot: Pause only on errors or ambiguous geometry. Verify autonomously otherwise.

Regardless of mode, the skills' checkpoint rules for complex features (fillets, revolves, polar arrays, fused geometry) still apply — but in zero-shot mode, treat them as autonomous verification points rather than user-facing pauses unless something looks wrong.

Setup

Dependency status is injected into context automatically at session start by a SessionStart hook. Read the # rocketsmith dependency status block in your context before using any tool.

  • If status: ready, proceed normally.
  • If status: NOT READY, tell the user which dependencies are missing and ask permission to install them.
  • Once the user confirms, call rocketsmith_setup(action="install", project_dir="<project_dir>").
  • Do not use cadsmith_* tools until all dependencies are ready.

Always pass project_dir when calling rocketsmith_setup. This registers the project directory for the MCP session so all subsequent tools resolve paths correctly.

Available MCP Tools

  • cadsmith_run_script — Execute a build123d Python script in an isolated uv environment (script_path, out_dir)
    • Runs the script with uv run --isolated --with build123d --with bd-warehouse so no host Python or conda env is required
    • The script must write one or more .step files to out_dir (which must exist)
    • Returns the list of STEP file paths produced by the script
    • Use this as the primary execution path — never call python, uv run, or conda run directly. They will fail or hit the wrong interpreter.
  • cadsmith_bd_warehouse_info — Instantiate a bd_warehouse parametric part and return its geometric properties and dimensional attributes (generator_class, generator_params, project_dir)
    • Use before writing Pass 2 scripts to discover clearance drill sizes, head diameters, nut thickness, and other dimensions needed for boolean cuts
    • Returns a Part model (bounding box, volume, source, generator metadata) plus an attributes dict with fastener-specific values (e.g. clearance_drill_sizes, nut_data, screw_data, head_diameter, thread_diameter)
    • Optionally exports a STEP file and Part JSON when project_dir is provided
    • Example: cadsmith_bd_warehouse_info(generator_class="HexNut", generator_params={"size": "M4-0.7", "fastener_type": "iso4032"})
  • cadsmith_generate_assets — Generate visualization assets for a STEP file (step_file_path, project_dir, outputs)
    • Always generates an STL mesh to gui/assets/stl/ for the 3D viewer.
    • outputs=["thumbnail", "gif", "ascii"] (default: all three). All three read from the STEP directly and run in parallel with STL generation.
      • thumbnail — single isometric PNG
      • gif — 360° rotating isometric GIF (rasterized, 36 frames at 10°/frame)
      • ascii — 360-frame text animation (1°/frame), used as a loading icon
    • Outputs written to gui/assets/png/, gui/assets/gif/, gui/assets/txt/ under the project directory.
    • Progress tracked per-part in gui/progress/<part_name>.json for the GUI.
  • gui_navigate — Navigate the GUI to a route path (e.g. /parts/nose_cone)
  • cadsmith_extract_part — Extract volume, bounding box, and centre of mass from a STEP file (step_file_path)
    • Use to verify dimensions numerically after visual inspection
  • openrocket_component (action="read") — Read the full component tree from an .ork design with mm-scaled CAD parameters (rocket_file_path, project_dir, no component_name)
    • Returns components (every length already in mm), derived (body_tube_id_mm, max_diameter_mm, motor mount block), and handoff_notes
    • You usually don't call this directly — the design-for-additive-manufacturing skill calls it when building the manifest. You only need it if you're verifying a feature value against the source design.
  • rocketsmith_setup — Check or install dependencies (action: check/install, project_dir: absolute path to the project)
    • Always pass project_dir

Read the full file on GitHub · 174 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 174 lines · 384 tokens per session scan A 9986874baab7

Subscribe to this mod's changes

cadsmith is an agent published in the GitHub repository ppak10/RocketSmith (21 stars, last pushed 4mo ago), licensed MIT. It adds 384 tokens to every session and 3,643 once invoked, about $0.0019 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other agents, from other repositories

skill-editor

Applies a single, minimal, generalized edit to a Logic-Lens skill (SKILL.md / guide / shared file) given a concrete failure diagnosis. Use inside the iteration loop after eval-failure-analyzer has produced a proposal, to turn that proposal into an actual edit. Mutates files; does NOT run evals or sync the cache — it…

hyhmrright/logic-lens · 84 tokens

code-reviewer

Expert code review specialist. Proactively reviews code for quality, security, and maintainability. Use immediately after writing or modifying code. MUST BE USED for all code changes.

Jamkris/everything-gemini-code · 38 tokens

planner

Expert planning specialist for complex features and refactoring. Use PROACTIVELY when users request feature implementation, architectural changes, or complex refactoring. Automatically activated for planning tasks.

Jamkris/everything-gemini-code · 37 tokens

refactor-cleaner

Dead code cleanup and consolidation specialist. Use PROACTIVELY for removing unused code, duplicates, and refactoring. Runs analysis tools (knip, depcheck, ts-prune) to identify dead code and safely removes it.

Jamkris/everything-gemini-code · 51 tokens

java-reviewer

Expert Java and Spring Boot code reviewer specializing in layered architecture, JPA patterns, security, and concurrency. Use for all Java code changes. MUST BE USED for Spring Boot projects.

Jamkris/everything-gemini-code · 40 tokens

cpp-build-resolver

C++ build, CMake, and compilation error resolution specialist. Fixes build errors, linker issues, and template errors with minimal changes. Use when C++ builds fail.

Jamkris/everything-gemini-code · 40 tokens