… 3.9.

- Expand test suite from 11 to 46 tests (92.55% coverage achieved)
- Add test_command_line.py with 11 CLI tests
- Add 15 new tests to test_autoplot.py
- Add 8 new tests to test_figure.py (scatter, log axes, subplots, etc.)
- Create vplot/__main__.py for module execution support
- Add .coveragerc for coverage configuration matching vspace
- Update GitHub Actions workflow with codecov integration
- Configure Python 3.9-3.12 testing matrix
- Add modernization plan to .claude/claude.md

Coverage breakdown:
- command_line.py: 100%
- colors.py: 100%
- __init__.py: 100%
- auto_plot.py: 98.39%
- figure.py: 90.05%
- Total: 92.55%

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <[email protected]>

Resolved conflicts in .github/workflows/tests.yml:
- Kept actions/setup-python@v5 (more standard than conda setup)
- Kept flexible test result publishing condition (runner.os == 'Linux')

No breaking changes to source code - only test expansions and coverage.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <[email protected]>

… GitHub Actions.

Changes:
- Use ubuntu-22.04 (explicit version) instead of ubuntu-latest
- Update to actions/checkout@v5 (from v4)
- Add diagnostic test step before full test suite
- Increase test timeout from 10 to 20 minutes
- Add -s flag to pytest for subprocess output visibility
- Update conditional checks from ubuntu-latest to ubuntu-22.04
- Standardize step names ("Install VPLanet", "Publish test results")
- Remove "Clone vplot repo" name (use default from vspace)

Testing limited to ubuntu-22.04 and Python 3.9 until GitHub Actions is debugged.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <[email protected]>

Clarified that vplanet.run() will either run the simulation or load
existing output depending on whether log files exist.

Note: Tests currently fail in CI due to "(null)" units in public vplanet
version. This is a known limitation already fixed in vplanet-private and
will be resolved when migrating to vplanet-private.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <[email protected]>

The diagnostic test was blocking the full test run from executing. Since
we know tests will fail due to the vplanet "(null)" units issue (a known
limitation in public vplanet), we should run all tests to see which ones
pass and get full coverage data.

Removing the diagnostic step allows the full test suite to run and provides
better visibility into which tests are affected by the vplanet issue versus
other potential problems.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <[email protected]>