BehalfID

One evaluated decision — allowed, denied, approval required, or error. Cached repeat decisions within a short window are not double-counted toward quota where the platform coalesces them.

Verification evaluation continues; enforcement does not degrade because of billing. Usage and plan limits are visible in the dashboard. Contact us if you need a higher committed volume.

Plans include an agent allowance and a billable seat limit derived from your workspace plan. Enterprise has no practical agent or seat cap under contract.

Three: Free and Pro are self-serve at checkout, Enterprise is via contact. That is the whole funnel — there is no fourth tier to wait for, and nothing is gated behind a sales call except Enterprise.

No. The policy engine, the decision semantics and every endpoint are identical on Free, Pro and Enterprise, and the request ceiling is 60 requests per minute per API key on all of them. Plans differ on monthly verification volume, log retention, webhooks, seats and organisational controls.

An account, an agent, and that agent's API key. Create the agent in the dashboard or with the CLI, then POST /api/verify with the agent, the action and the resource — the response tells you whether to proceed. The quickstart walks the whole loop end to end.

The TypeScript SDK and the CLI are first-party, and the CLI installs action-time hooks for coding agents such as Claude Code, Codex and Cursor. There is also an MCP server and the Action Gateway. Anything that can make an HTTPS request can call the API directly — the SDK is a convenience, not a requirement.

The workspace returns to the Free entitlements. Existing agents and policies are not deleted and verification keeps working — the free limits on volume, retention and seats apply again, creating further agents is blocked past the free allowance, and webhook delivery is disabled unless the workspace still qualifies for it.