Management commands¶
| Surface | Purpose | Inspection example |
|---|---|---|
setup |
Discover local service endpoints without model installation | ribbit setup --json |
doctor |
Diagnose configuration, picker, extensions, and explicit provider probes | ribbit doctor --json |
providers |
Add, list, and remove endpoint configurations | ribbit providers list --json |
models |
List models from an explicitly selected provider | ribbit models list --provider local |
profiles |
Set, inspect, list, and remove named inference settings | ribbit profiles list --json |
route |
Inspect the route for a command | ribbit route inspect ask --profile local-small --json |
commands |
List and describe commands; validate named definitions | ribbit commands list --json |
types |
Inspect built-in and installed type contracts | ribbit types describe @ribbit/summarize --json |
extensions |
Scaffold, check, test, add, list, and remove trusted extensions | ribbit extensions list --json |
run |
Invoke a named command with argument overrides | ribbit run brief --file meeting.txt --profile local-small |
flow |
Validate, plan, or run saved and inline flows | ribbit flow plan examples/flows/triage.yaml |
init |
Create project config and optionally append agent guidance | ribbit init --agent codex |
completions |
Emit Bash, Zsh, or Fish completions | ribbit completions bash |
Examples that refer to a profile, definition, or file require it to exist. models list and doctor --probe perform explicit network requests. Catalog/help/completion/flow planning inspect declarations without importing installed extension code.
init --agent supports codex, claude, cursor, and opencode. Initialization preserves existing unrelated guidance and does not replace an existing project config. Generated completions include named definitions.
See model setup, extension authoring, and named definitions for task-specific instructions.