Set up Skene configuration for a project. Use when starting a new PLG analysis, when .skene.config is missing, or when the user says "skene init", "set up skene", or "configure skene".
Set up Skene automatically. The user runs /skene-init, everything gets configured, and they see a one-line summary. Zero questions unless something is genuinely missing and cannot be inferred.
Config reference: Configuration guide.
Make init invisible. Detect everything from existing config, env vars, and user-level config. Only ask the user if the agent truly cannot figure it out (e.g. API key missing everywhere).
Discover CLI options
Run uvx skene config --help to see what flags are available. Use only flags that appear in the output.
Check what already exists (run in parallel where possible):
ls .skene.config 2>/dev/null — project configls ~/.config/skene/config 2>/dev/null — user-level configecho n | uvx skene config --show — full resolved config (piped to avoid blocking)printenv SKENE_API_KEY SKENE_PROVIDER 2>/dev/null — env varsDecide what to do (no user interaction needed):
Case A — Config exists and is valid (provider set, API key present, provider/model match): Done. Report summary.
Case B — Config exists but has issues (provider/model mismatch, missing API key, etc.): Fix what you can automatically:
.skene.config with correct defaults.SKENE_API_KEY in env → fine, note it.SKENE_API_KEY in your environment, then re-run /skene-init."Case C — No config at all:
uvx skene config --init (check --help first to confirm this flag exists)..skene.config TOML directly.openai.echo n | uvx skene config --show.Final validation
echo n | uvx skene config --show
Confirm provider, model, API key. If good, report.
Report result
"Skene init complete. Provider: X, model: Y. Run /skene-analyze next."
The user CANNOT type into agent-spawned terminals. The agent handles ALL input.
echo n to uvx skene config --show to skip the edit prompt.uvx skene config (no flags) — multi-step interactive. Write .skene.config directly..skene.config manually. If uvx itself can't run, tell user to run pip install uv in their terminal.SKENE_API_KEY.One line: "Skene init complete. Provider: X, model: Y. Run /skene-analyze next."