steward-cli 0.5.0

Steward — aligns and maintains resident agents across Culture projects.

steward

Steward aligns and maintains resident agents across Culture projects.

Install

pip install steward-cli

Or, with uv:

uv tool install steward-cli

Usage

steward --version
steward --help

# Show a Culture agent's full configuration in one view
# (CLAUDE.md + the parallel culture.yaml + .claude/skills/ index).
# Run from inside a Steward checkout — the command finds the agent-config
# skill script via a walk-up from the current working directory.
steward show ../culture
steward show ../daria

steward show is a thin wrapper over the agent-config skill at .claude/skills/agent-config/scripts/show.sh. The skill remains the canonical implementation; the CLI is the typed entry point.

# Diagnose a single sibling repo against the AgentCulture sibling pattern
# (portability + skills-convention checks). Exits non-zero on findings.
steward doctor ../culture
steward doctor ../culture --json --check portability

# Walk every culture.yaml in the workspace, score each declared agent
# against the corpus baseline, and write per-repo feedback into
# <target>/docs/steward/steward-suggestions.md. Diagnostic-only.
steward doctor --scope siblings

steward doctor is read-only diagnosis. --scope self (default) runs the single-repo invariant checks against TARGET. --scope siblings walks every culture.yaml in the workspace, refreshes docs/perfect-patient.md from the corpus, and writes per-target feedback into each sibling's docs/steward/steward-suggestions.md. The repair mode (--apply) is on the roadmap but not implemented yet.

See docs/sibling-pattern.md for the contract doctor honors and docs/perfect-patient.md for the auto-generated corpus baseline.

See CLAUDE.md for project-shape, build/test/publish details, and the skills convention.

License

MIT — see LICENSE.