Problem
docs/commands.md only documents ~15 commands but the CLI surface has 50+. Agents and operators reading this reference file have no discovery path for most commands.
Missing command groups (not mentioned at all)
| Group |
Commands |
| Observability |
squads log, squads obs, squads usage, squads cost, squads budget |
| Automation |
squads trigger, squads autonomous, squads approval |
| Execution |
squads orchestrate, squads eval, squads context (aka feed) |
| Operations |
squads kill, squads runs, squads credentials, squads services, squads tier, squads brief |
| Verification (P0) |
squads contract |
| Knowledge |
squads goals (distinct from squads goal), squads review, squads catalog, squads release |
| System |
squads providers, squads update, squads login/logout/whoami |
Current docs/commands.md scope
The file documents a "For Humans" section (init, add, doctor, run, autopilot, status, sessions, dash, goal, results, stats) and a "For Agents" section (env show/prompt, status, memory, feedback, exec, kpi). Both sections miss most of the surface.
Impact
Agents calling squads commands --json get the full live tree; agents reading docs/commands.md get a ~30% sample. The gap is most painful for:
- Agents doing observability work (
obs, log, usage)
- Agents managing background runs (
kill, runs)
- Agent contract / verification work (
contract)
Fix
- Expand "For Humans" section with observability, automation, and operations groups
- Expand "For Agents" section with verification, context, and catalog commands
- Add each group's primary flags and a one-line example
- Note that
squads commands --json is the always-current machine reference
Source: docs-eval agent
Problem
docs/commands.mdonly documents ~15 commands but the CLI surface has 50+. Agents and operators reading this reference file have no discovery path for most commands.Missing command groups (not mentioned at all)
squads log,squads obs,squads usage,squads cost,squads budgetsquads trigger,squads autonomous,squads approvalsquads orchestrate,squads eval,squads context(akafeed)squads kill,squads runs,squads credentials,squads services,squads tier,squads briefsquads contractsquads goals(distinct fromsquads goal),squads review,squads catalog,squads releasesquads providers,squads update,squads login/logout/whoamiCurrent docs/commands.md scope
The file documents a "For Humans" section (init, add, doctor, run, autopilot, status, sessions, dash, goal, results, stats) and a "For Agents" section (env show/prompt, status, memory, feedback, exec, kpi). Both sections miss most of the surface.
Impact
Agents calling
squads commands --jsonget the full live tree; agents readingdocs/commands.mdget a ~30% sample. The gap is most painful for:obs,log,usage)kill,runs)contract)Fix
squads commands --jsonis the always-current machine referenceSource: docs-eval agent