Getting started
Plans & access
The access ladder
| Grow | Scale | Custom | |
|---|---|---|---|
| Active API keys | 1 | 2 | 10 |
| Read every endpoint | ✓ | ✓ | ✓ |
| MCP server | ✓ | ✓ | ✓ |
| Writes: prompts, content states, queue tasks | — | ✓ | ✓ |
| Generate drafts (spends credits) | — | ✓ | ✓ |
| Bulk exports and webhooks | — | — | ✓ |
| Rate limit | 60/min | 300/min | 600/min |
| Connected AI apps (MCP sign-in) | 5 per member, where sign-in connections are enabled | ||
Solo and Track don’t include API or MCP access — the Developers page won’t mint a key there, and a key from a workspace that has moved down answers 402 plan_required. See pricing for what each plan measures.
What switches access on
Two conditions, both checked on every single call:
- the workspace’s plan is Grow, Scale or Custom —
402 plan_requiredotherwise, and - the workspace is approved —
403 workspace_not_approvedwhile it is pending or was declined.
Read Get plan and usage to see which plan a key is operating under, what this month has spent, and every limit that applies. A null limit means unlimited on that plan.
Several workspaces on one plan
Larger customers buy one plan for several brands. Where that is the case, billing_scope is account, and the allowances — prompts, answers, credits and the rate limit — are shared across all of the workspaces on it, not granted to each. Two integrations pulling hard at the same time will feel each other.
What the API deliberately can’t do
The API offers what the dashboard offers, and no more. These are missing on purpose, not yet to be built:
- Deleting prompts. Deleting one takes its measurement history with it. Archive instead — it frees the slot and keeps the record.
- Refining a draft. Refinement is an editing conversation, and it belongs where the editor can see the result.
- Publishing to a CMS. Publishing is a decision with consequences on your own site.
- Running probes on demand.Sweeps run on the plan’s cadence; an on-demand run would spend your answer allowance from a script.