Early adopter setup

python3.12 --version
pipx install --python python3.12 "git+https://github.com/codemower-ai/code-mower.git@v0.5.0-alpha.8"
code-mower --version

code-mower init --easy
code-mower init --easy --apply --output-dir .code-mower.generated

code-mower doctor --preflight --json

--v05 is the equivalent versioned preset for scripts. Fix failures first. Treat warnings as setup guidance, especially for private repositories and paid Actions minutes.

If you want to prove the whole first-user path before touching a real repository, run the package-install rehearsal. It installs Code Mower into a clean virtual environment, creates a toy Git repo, runs easy-mode setup, runs doctor, generates a value report, and proves cloud upload/dogfood paths stay dry-run.

code-mower migration package-install-rehearsal \
  --package-spec "git+https://github.com/codemower-ai/code-mower.git@v0.5.0-alpha.8" \
  --python "$(command -v python3.12)" \
  --json

The rehearsal leaves inspectable artifacts: calibration evidence, reviewer metrics, lane policy, value report, cloud export bundle, upload dry run, and CodeMower.com dogfood dry run.

For a real pilot repository, generate the starter value report:

code-mower calibration value-report .code-mower.generated/calibration-corpus.json \
  --output .code-mower/reviewer-value-report.md

Replace the starter corpus with known-clean and known-blocked PRs before promoting any lane to a merge gate.

Sign in, create or join a team, issue a team token, then store it with the local CLI. Upload remains dry-run first and metadata-only by default. CodeMower.com shows private team signal today; cross-team cohort benchmarks are still launch-readiness work. Team metadata export is available to team members, and uploaded metadata deletion is available to team owners/admins.

code-mower cloud setup --token-stdin \
  --team-id YOUR_TEAM_SLUG \
  --install-id your-laptop \
  --out ~/.config/code-mower/tokens/your-laptop.env

source ~/.config/code-mower/tokens/your-laptop.env
code-mower cloud dogfood --json
code-mower cloud dogfood --yes --json

Use the dry run to inspect the exact metadata shape first. The confirmed dogfood command is what makes the dashboard useful over time: uploads, repo rollups, provider/lens signal, cost, latency, and recommendations.

Sign in to CodeMower.com or inspect what data can upload.

• Use one low-risk repository first.
• Keep reviewer lanes manual.
• Do not add recurring schedules.
• Do not put hosted reviewers into branch protection yet.
• Promote lanes only after local calibration data supports it.

Public setup docs live in docs/first-user-install-rehearsal.md. The shorter walkthrough is docs/try-in-10-minutes.md.