Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
66 lines (49 loc) · 2.7 KB

CONTRIBUTING.md

File metadata and controls

66 lines (49 loc) · 2.7 KB
Raw

Contributing

Read AGENTS.md first — it defines the trust boundary this repo protects (secrets stay local, the proof claim stays narrow, artifact sets stay coherent). This file only covers the mechanics.

One-time setup

pnpm install                 # repo root: installs Biome + lefthook, registers git hooks
bash scripts/bootstrap-vendor.sh   # required: vendors gnark with the reviewed local patches

Never run plain go mod vendor; it drops the hand-applied patch. Use the bootstrap script, and scripts/check-vendor-drift.sh to verify.

Optional but recommended (CI enforces it either way):

go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.12.2

Everyday commands

Command What it does
pnpm lint / pnpm lint:fix Biome lint + format check (fix) for all TS/JS packages
golangci-lint run ./... Go linting (gofmt, staticcheck, errcheck, …)
pnpm test:all Full aggregate gate — see below
pnpm test:fast Same, minus the slow circuit/full-proof/contract suites

Git hooks (via lefthook) format and vet staged files on commit and run the repo-wide linters on push.

The aggregate test gate

pnpm test:all (= scripts/test-all.sh) is the "can we ship this?" command. A full green run covers:

  • Go engine: build, vet, lint, every package test, including the real Groth16 ownership/multi/destination round-trips with tamper cases (PROOF_TOOL_RUN_FULL_PROOF=1).
  • WASM prover + MSM worker builds (GOOS=js GOARCH=wasm).
  • packages/client-ts: build + tests (CIP-3 golden vectors, worker protocol).
  • Web app: typecheck, all vitest suites (components, claim/reclaim libs, claim API route contracts, real-CBOR claim tx building, e2e-harness units), plus verify:proof-release — the signed proof-asset coherence check.
  • Desktop app: typecheck + tests; the three Rust crates when cargo exists.
  • Plutus validators: cabal test ownership-verifier-test — the full suite running the compiled validators with real proof fixtures and negative paths — when cabal exists.

The authoritative section list lives in the header of scripts/test-all.sh; if this summary and the script disagree, the script wins.

Sections whose tools are missing are reported as SKIP, not silently dropped. CI (.github/workflows/ci.yml) runs every section on each PR.

What test-all does not cover (needs credentials/hardware): the live preprod claim journey (pnpm --dir apps/ownership-proof-web test:e2e:preprod) and real browser-WASM proving in an actual browser.

Pull requests

CI must be green. The PR template checklist mirrors the AGENTS.md risk model — take it seriously, especially the artifact-coherence and no-secrets items.