vcad - Moses Yebei

The Problem

Traditional CAD kernels are monolithic, proprietary, and difficult to embed in modern web or AI-driven workflows. vcad addresses this by providing an open-source, parametric BRep (Boundary Representation) kernel written in Rust, designed for the AI era—meaning it is built to be controlled programmatically, integrated with AI agents, and run in browsers or on desktops.

What This Does

vcad is a full-featured CAD kernel and application suite. At its core, the crates/ directory contains a modular Rust workspace: vcad-app handles the document model and feature operations (extrude, revolve, boolean ops), vcad-crdt provides conflict-free replicated data types for collaborative editing, and stepperoni is a parser/lexer for a scripting language. The vcad-cli crate offers a terminal UI and a vcad command-line tool for export/import operations. A vcad-desktop crate is a Tauri v2 shell for a native desktop app.

The project also includes a web application (React/Three.js), a Supabase backend for cloud sync and authentication, and an MCP server (@vcad/mcp) that exposes CAD operations to AI agents. The changelog/ directory shows rapid feature development, including STEP import/export, physics simulation with Rapier3D, and AI-assisted text-to-CAD.

How To Use It

Setup: The project is a Rust workspace managed with Cargo. It has a critical external dependency: the tang math workspace must be cloned as a sibling directory (../tang). The web app uses npm.

Configuration: The supabase/ directory contains database migrations, and the crates/vcad-chat/src/auth.rs file handles authentication. Environment variables for Supabase keys are likely required for cloud features, but the exact names are not documented in the provided files.

Running it: Clone the required sibling dependency first git clone git@github.com:ecto/tang.git ../tang

Run Rust tests and lints

cargo test --workspace cargo clippy --workspace -- -D warnings

Run the web app locally

npm ci npm run dev -w @vcad/app

Launch the desktop app (requires Tauri system deps)

npm run tauri:dev

The CLI entry point is crates/vcad-cli/src/main.rs, and the MCP server is installed via npm install -g @vcad/mcp.

Real-World Use

A practical scenario is an AI-driven design automation pipeline. An agent uses the MCP server to create a parametric part, the kernel evaluates the feature tree, and the result is exported for manufacturing.

use vcadkernel::Solid;

let solid = Solid::cube(100.0, 60.0, 20.0); let hole = Solid::cylinder(10.0, 25.0, 32); let result = solid - hole; let mesh = result.tomesh(32);

Code Health & Issues

High - Single point of failure in build: The hard dependency on a sibling ../tang workspace is fragile. A fresh clone without that specific setup will fail to compile. This is documented but is a significant onboarding hurdle. High - Missing core crate: The README references crates/vcad-kernel as the unified API, but this directory is not present in the provided file tree. The actual structure uses vcad-app and others. This is a documentation and structural inconsistency. Med - Notarization gap: The desktop app is unsigned for macOS, requiring manual workarounds for users. This is a release-quality issue. Med - Rapid, unverified changelog: The changelog/ has entries dated into 2026, suggesting an aggressive release cadence. Without seeing the CI logs, it's unclear if all these features are tested. The repo does have a ci.yml and some test files, which is a positive sign. Low - Mixed language stack: The presence of cad-lib/ with .loon files (a different language) alongside Rust and TypeScript adds complexity.

The Bottom Line

vcad is an ambitious and technically interesting project that successfully brings a modern CAD kernel to the Rust/WASM ecosystem. It is well-structured with a modular crate layout and has active CI. However, the fragile sibling dependency, missing vcad-kernel crate, and rapid release cadence suggest it is not yet production-stable. It is best suited for developers and teams who want to experiment with programmatic, AI-driven CAD and are comfortable managing a complex build environment.

What the analyser found

Deployment readiness

7/7

Container image
CI pipeline
Lockfile committed
Test suite
README
License
No committed secrets

Composition

4,119 files

JSON1136
Rust359
TypeScript317
TSX240
Markdown110
TOML56

ReactNext.jsFlaskExpressSpringDocker

Module dependencies

The 10 most depended-upon modules of 582, from static import analysis. Red outline marks a module in an import cycle.

graph LR n0_raining_src_generators_types["types"] n1_raining_src_generators_utils["utils"] n2_packages_auth_src_client["client"] n3_s_auth_src_stores_auth_store["auth-store"] n4__app_src_data_examples_index["index"] n5_packages_core_src_types["types"] n6_mecheval_harness_src_task["task"] n7_ectronics_pcb3d_pcb_geometry["pcb-geometry"] n8_packages_wasmosis_src_types["types"] n9_mecheval_harness_src_solver["solver"] n9_mecheval_harness_src_solver --> n6_mecheval_harness_src_task n1_raining_src_generators_utils --> n0_raining_src_generators_types classDef cyc stroke:#ef4444,stroke-width:2px; class n4__app_src_data_examples_index,n9_mecheval_harness_src_solver cyc;

What would stop me shipping this

Ranked by severity × confidence × production reach. Reach is the honest discriminator across a collection that is mostly other people's code: the same finding matters more in something that ships.

critical1

high2

medium3

low2

Keep secrets out of workflows a fork can triggercritical

.github/workflows/fix-ci.yml

ANTHROPIC_API_KEY

Contributor-controlled input reaches a job holding a publish token or cloud credential, so a pull request becomes credential exfiltration.

Fix: Move secret-using steps into a workflow_run job that never checks out pull request code, or gate on an environment with required reviewers.

Pin third-party GitHub Actions to a commit SHAhigh4 occurrences

.github/workflows

dtolnay/rust-toolchain@master, Swatinem/rust-cache@v2, anthropics/claude-code-action@v1, tauri-apps/tauri-action@v0

A tag can be moved, so the action running with your token and secrets is whatever its owner last pushed; this is how tj-actions/changed-files leaked secrets from thousands of repos.

Fix: Replace each @vN with the 40-character commit SHA, keep # vN as a comment, and let Dependabot bump the SHAs.

Stop discarding the exit code of steps whose failure mattershigh2 occurrences

.github/workflows/ci.yml

lines 225, 239 discard failure

The job reports green while the artifact it was supposed to produce was never produced, so a crash becomes a successful run built on stale data.

Fix: Let the step fail, or test the specific expected condition instead of discarding all exit codes.

Pin the container base image by digestmedium

packages/hf-space/render-api/Dockerfile

node:20-slim

An untagged or mutable base means today's build and last month's contain different libc and a different CVE set, with no record of which shipped.

Fix: Use image:tag@sha256:<digest> and enable Dependabot's docker ecosystem.

Move large binaries to Git LFS or out of the repositorymedium

packages/kernel-wasm/vcad_kernel_wasm_bg.wasm

1 blobs over 5MB: vcad_kernel_wasm_bg.wasm 9.5MB

One repository here carries twenty blobs over 5MB including a 10.9MB spreadsheet, so every clone and every CI checkout pays for data nobody diffs.

Fix: Track those extensions with LFS, or move datasets to object storage and fetch them in a setup step.

Set persist-credentials: false on checkoutmedium

.github/workflows/agent.yml

checkout keeps the token, then dependencies are installed

The token stays in .git/config for every later step, so a malicious postinstall script reads a pushable credential without one ever being passed to it.

Fix: Add with: persist-credentials: false, and pass an explicit token only to the step that pushes.

Set timeout-minutes on the workflow jobslow4 occurrences

.github/workflows/agent.yml

4 workflow(s) declare no job timeout

A wedged step runs to the six-hour platform default, which on a two-hourly schedule means three runs overlap behind it.

Fix: Add timeout-minutes with a realistic bound to each job.

Add the repository convention files this project lackslow3 occurrences

missing .editorconfig, .gitattributes, a formatter config

Without them one contributor's editor writes tabs into a Python file, a shell script commits with CRLF and fails in the container, and a notebook diff is unreviewable.

Fix: Add .editorconfig, .gitattributes with text=auto eol=lf, and a formatter config.

Checked deterministically against the repository tree and a bounded set of its files: committed credentials, unpinned actions and base images, missing lockfiles and update bots, workflows that discard failures, published advisories against the declared dependencies, runtime configuration, licensing and notebook reproducibility. No language model is involved in this section.