when need python, use uv run python
jupyter-jcliversion must have a single source of truth:[project].versioninpyproject.toml.- Do not leave a hard-coded duplicate version in Python modules or docs.
- After bumping
[project].version, runuv lockto refreshuv.lockso the editablejupyter-jclientry matches the new release version; do not hand-edituv.lock. - When bumping a release version, update any user-facing docs that mention the current released version or version verification workflow.
- After a version bump, verify both:
uv run python -m jupyter_jcli --versionj-cli --version
- If a test asserts version output, make it compare against installed distribution metadata instead of a copied literal.