Approval gate
A hold that pauses a high-risk action until a designated human approves it.
category ▸ Governance
Approval gate, in plain language.
An approval gate is the human-in-the-loop control: when a proposed action crosses a risk threshold, the runtime suspends it in a pending state and routes it to a person with the authority to approve or deny. Execution only resumes on an explicit, recorded decision.
Approval gates are how regulated workflows keep a human accountable for consequential outcomes — a payout, a client communication, a filed return — without abandoning automation everywhere else. The gate is selective: low-risk work flows, high-risk work waits.
How Cortex implements it.
This term isn't abstract here — it maps to a real capability in the runtime. Here is exactly how Cortex enforces or relates to it.
In Cortex, an action that is high-risk or marked requires-approval lands pending_approval and surfaces in an approval inbox. Approve executes it; deny records the decision and it never executes. The hold returns a 409 verdict with the approver's receipt recorded.
Oversight modes decide which actions reach the gate, and the action's own risk policy sets a floor the gate always honors — so the approval requirement cannot be configured away.
Keep building the vocabulary.
These terms sit next to this one in the governed-AI model — follow the thread to see how the controls connect.
Oversight modes
Per-agent autonomy levels that decide whether an action executes now or waits for a human.
Break-glass
An audited override that lets an authorized human force a held action through, with a recorded reason.
Action Fabric
A lifecycle that makes every agent action a governed object: dry-run, propose, approve, execute, compensate.
Governance gate
A single decision point that allows, holds, or blocks a proposed agent action.
See Approval gate enforced, not just defined.
Join the waitlist and watch the controls in this glossary return real verdicts, seal real evidence, and trace every fact back to its source.