---

name: development-maintenance description: "Make safe Nilearn checkout code, test, documentation, changelog, and maintenance changes without violating repository architecture or conventions." disable-model-invocation: true metadata: disco-role: operating license: BSD 3-Clause

Nilearn Development Maintenance Skill

Use this sub-skill when working inside the Nilearn source checkout on code, tests, documentation, changelog entries, maintenance scripts, or repository configuration. It is for coding-agent maintenance tasks, not user-facing neuroimaging analysis workflows.

Route Tasks

• Code changes: Follow style, public API, backcompat, deprecation, and estimator conventions in development guide.
• Import architecture: Check the import-linter layer rules before adding cross-package imports. See development guide and troubleshooting.
• Tests and docs: Pick narrow no-network tests first, mark generated or modified tests with @pytest.mark.ai_generated, and use doc/gallery guidance in testing and docs.
• Validation planning: Use the bundled helper to suggest targeted commands without running them: python scripts/select_safe_tests.py <changed-paths> from this sub-skill directory.
• Failures: Diagnose import-linter, optional plotting, baseline image, network/data skip, estimator-check, and pre-commit failures with troubleshooting.

Quick Checklist

1. Keep changes focused; do not add dependencies or unrelated refactors.
2. Respect pyproject.toml ruff settings: double quotes, 4-space indents, and 79-character Python lines.
3. Use absolute internal imports and obey Nilearn's layered import contracts.
4. Preserve scikit-learn-style estimator APIs: fit returns self, learned attributes end with _, and fit overwrites fitted state.
5. Add or update small synthetic tests near the changed module and mark any generated or modified tests with @pytest.mark.ai_generated.
6. Add an entry to doc/changes/latest.rst for PR-worthy user, API, code, test, doc, plotting, maintenance, or deprecation changes.

Boundaries

• Route analysis usage, Niimg operations, GLM, maskers, decoding, connectivity, plotting, or surface workflows to their domain sub-skills unless the task is specifically about changing Nilearn source/tests/docs.
• Do not run broad test suites, slow examples, visual baseline generation, network dataset downloads, or native data fetchers unless the user explicitly asks and the environment is appropriate.
• Do not create Codex-, Claude-, or export-specific files from this sub-skill.

References

• Development guide
• Testing and docs
• Troubleshooting
• Safe test selector