Thanks for your interest in tensorlake-docai. PRs are welcome.
Maintainer bandwidth is limited — response times are best-effort, often a few business days. To keep things moving:
- Small, focused PRs land fastest. One change per PR.
- Big refactors or new model integrations: please open an issue first to align on the design before writing code.
git clone https://github.com/tensorlakeai/openingest
cd openingest
pip install -e ".[dev]"
cp .env.example .env # fill in only the keys you needRun the smoke tests:
pytest tests/ -qFormat before committing:
black src/ tests/ examples/
ruff check src/ tests/ examples/-
pytest tests/ -qpasses - New code has at least one test (mock external API clients)
- No new required env vars without a
.env.exampleupdate + README mention - No hardcoded keys, endpoints, or bucket names
- If you added a new OCR backend, document it in
docs/models.md
- Adding a new ML provider integration without ongoing maintenance commitment
- Large architectural rewrites — the pipeline is intentionally simple
- Dependency bumps that pull in heavy new transitive deps
- PRs that mix unrelated changes
Open a GitHub issue with: version (or commit SHA), what you ran, what you expected, what happened, and a minimal reproducer.
For security issues, see SECURITY.md — don't file public
issues.
By contributing you agree your contributions are licensed under Apache-2.0, the project license.