Skip to content
Developer Guide
Contributing

Contributing

DataKit is a Go monorepo, so contributions are expected to include code, tests, and documentation updates when behavior changes. Use a short-lived branch, validate locally, and open a pull request with a conventional title.

    flowchart LR
  I[Pick issue or change] --> B[Create branch]
  B --> C[Implement change]
  C --> T[Run lint and tests]
  T --> P[Open pull request]
  P --> R[Address review]

Branch conventions

Create branches from your working branch using the conventions documented in CONTRIBUTING.md:

git checkout -b feature/your-feature-name
# or
git checkout -b fix/issue-number-description

Keep changes scoped to a single feature or fix so the resulting PR is easy to review.

Pull request workflow

Before opening a PR:

  1. run the relevant tests
  2. run linting
  3. update docs when user-facing behavior changes
  4. make sure generated artifacts were regenerated instead of edited by hand

The root guide also expects PR titles and commit messages to follow conventional commit style, for example:

  • feat: add PII validation to lint command
  • fix: resolve binding reference validation
  • docs: update CLI reference documentation

Code style

For Go code, follow the repository guidance from CONTRIBUTING.md:

  • use standard Go formatting tools such as go fmt ./...
  • keep exported identifiers documented
  • write lowercase error messages without trailing punctuation
  • prefer table-driven tests for multiple scenarios
  • mock external dependencies in unit tests

Local validation

The repository-level targets are the default pre-PR checks:

make lint
make test-unit
make test-integration
make test-e2e

If you only changed one module, you can still run focused go test commands, but the shared Make targets are the canonical project checks.

Documentation and generated files

Update relevant docs alongside behavior changes. For generated files such as CRDs, deepcopy files, protobuf outputs, or scaffolded manifests, change the generator input and re-run the generator instead of hand-editing generated output.

Next Steps

Testing