The repository-docs standard requires every SDK repo to have an AGENTS.md at the repository root. sentry-python currently has no AGENTS.md, meaning AI agents working in this repo have no architecture context, no testing conventions, no build instructions, and no SDK-specific standard overrides. The file that makes agentic contributions accurate and consistent.
What needs to be done
- Create
AGENTS.md at the repo root. Use agents-md skill.
- Create
CLAUDE.md as a symlink: ln -s AGENTS.md CLAUDE.md
- Fill in all SDK-specific sections:
- Package manager and key commands (pip/tox/make:
tox -e linters, tox -e py3.12, make test)
- Architecture overview (core SDK in
sentry_sdk/, integrations in sentry_sdk/integrations/, tests in tests/)
- Development workflow (tox environments, linting via
tox -e linters, integration test matrix)
- Testing conventions (pytest, test file naming
test_*.py, tox environments per integration, tests/integrations/ layout)
- Standards overrides (any known deviations from the SDK standards)
- Evaluate adding a extra AGENTS.md or similar file for AI integrations.
- Verify accuracy: have an AI tool attempt a small contribution using the file as context before merging
The repository-docs standard requires every SDK repo to have an
AGENTS.mdat the repository root.sentry-pythoncurrently has noAGENTS.md, meaning AI agents working in this repo have no architecture context, no testing conventions, no build instructions, and no SDK-specific standard overrides. The file that makes agentic contributions accurate and consistent.What needs to be done
AGENTS.mdat the repo root. Use agents-md skill.CLAUDE.mdas a symlink:ln -s AGENTS.md CLAUDE.mdtox -e linters,tox -e py3.12,make test)sentry_sdk/, integrations insentry_sdk/integrations/, tests intests/)tox -e linters, integration test matrix)test_*.py, tox environments per integration,tests/integrations/layout)