Operational limits

The caps emisar enforces so one noisy action, one busy agent, or one aggressive export can't overwhelm a page or the API. Most are generous; these are the ones worth knowing before you build against them.

What	Cap	At the cap
Action output	Set per action in the pack (a few KB to a few MB)	Bytes past the cap are dropped from the delivered output but counted, and the run is flagged truncated.
Live output in the dashboard	The most recent 500 progress lines	Older lines are trimmed from the browser view with a note; the run history and the host journal keep the full record.
MCP list pages	15 per page by default (up to 50; `recent_runs` up to 100, `find_actions` up to 15)	Continue with the cursor the previous response returned; a catalog cursor is bound to its exact query and expires after 15 minutes.
MCP long-poll wait	Up to 60 seconds per call	`run_action`, `wait_for_run`, and `recent_runs` return the current state at the deadline; one credential may hold 8 concurrent waits per node.
MCP response size	512 KiB per tool result	Oversized structured output is dropped and replaced with a `wait_for_run` continuation that fetches it.
Runbooks over MCP	Up to 32 steps, 16 runners per step, 256 runs, and a 56 KiB definition	Larger runbooks still run from the dashboard; only the model-facing projection is bounded.
Audit export	100 events per page, 1,000 maximum	Cursor-paginated forward; page with the cursor each response returns.
Audit retention	7 days on Free, 90 on Team, 365 on Enterprise	Fixed per event when it's written; stream to your SIEM to keep it longer.

Scripting against the API?

The full MCP contract — every method, parameter, error code, and the idempotency rules that make a retried dispatch safe — is in the MCP reference.

← Previous

Security model