Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
48 lines (35 loc) · 1.68 KB

CONTRIBUTING.md

File metadata and controls

48 lines (35 loc) · 1.68 KB
Raw

Contributing To Rula

Rula is a local-first realtime voice-avatar runtime. Contributions are welcome when they keep the project secure, reproducible, and honest about demo vs production readiness.

Before You Open A PR

  • Keep secrets, model weights, generated media, logs, SQLite databases, and local operator notes out of Git.
  • Do not add cloud inference to the default runtime path.
  • Preserve the realtime hot-path envelope: session_id, turn_id, generation_id, branch_state, seq, and optional pts_ms.
  • Drop stale generation_id artifacts silently at every hop.
  • Keep the runtime a modular monolith unless there is measured scale pressure.
  • Prefer small typed contracts over framework magic.

Development Checks

Run the focused checks before submitting changes:

$env:PYTHONPATH=(Resolve-Path ".\apps\agent\src").Path
python -m unittest discover apps\agent\tests
npm run protocol:test
npm run web:test
npm run web:build

For local Docker validation:

cd infra/wsl
docker compose config --quiet

Pull Request Expectations

  • Explain the user-visible behavior or operational risk being changed.
  • Include tests for protocol, turn-state, interruption, API, or UI behavior when touched.
  • Keep unrelated refactors out of the PR.
  • Call out model, license, GPU, latency, or air-gap implications explicitly.
  • Do not paste tokens, private paths, hostnames, SSH config, or local deployment runbooks.

Good First Areas

  • README and release evidence improvements.
  • Local eval scripts and acceptance reporting.
  • Protocol tests around stale-generation dropping.
  • Browser-side audio/face sync diagnostics.
  • Documentation for model licensing and attribution.