Before dispatching several agents on one mission, define who owns each write lane, who acts as the independent verifier, and what budget ceiling binds the run. The team planner converts three inputs into a set of role profiles you can hand to your own runtime.

## Request a team plan via HTTP [#request-a-team-plan-via-http]

```bash
curl https://starlightintelligence.ai/api/v1/team-plan \
  -H "content-type: application/json" \
  -d '{"teamSize":4,"maxConcurrent":2,"budgetUsd":20}'
```

| Field | Allowed values & meaning |
|---|---|
| `teamSize` | `2`, `4`, `8`, `12`, or `144` role profiles |
| `maxConcurrent` | Maximum parallel active lanes; capped at `12` |
| `budgetUsd` | Shared mission budget assumption in USD |

## Authority & safety contract [#authority--safety-contract]

- The response carries `status: "planning-only"`.
- It makes zero model calls, stores no customer data, and requires no credentials.
- Request payloads are capped at 4 KB.
- Prefer interactive browser controls? Open `https://starlightintelligence.ai/agent-kits/teams` to configure the same profile set visually.