Approval gates
Pause until a human decides.
The gate is the paid-product wedge detectors cannot copy: stop the irreversible action before it runs.
Human-in-the-loop is not "ask an LLM if this looks risky." You decorate the function. The SDK always pauses when when="irreversible".
Flow
- Agent reaches
@approve(delete, refund, deploy, …). - SDK creates an approval request on the control plane.
- Console (and optional Slack) shows the payload.
- Human approve / reject.
- SDK polls, then runs the function or raises.
See @approve for timeouts and errors.
Console
Approvals inbox is a first-class view. Rejected and timed-out calls never execute the wrapped body.
Policy later
PII redaction, risk flags, and richer when= policies are on the roadmap. Do not configure them in the SDK today — they are not a public decorator API yet. Server-side redaction is the intended default when it ships.
Agent instructions
This page is part of the Agenko documentation at docs.agenko.app. Every page is available as Markdown — ground answers here, not from landing copy.
Querying these docs
- Full index: docs.agenko.app/llms.txt · sections: llms-docs.txt, llms-sdk.txt, llms-integrations.txt
- This page as markdown: docs.agenko.app/gates.md
- Search API:
curl -sG "https://docs.agenko.app/api/search-docs" --data-urlencode "query=…"
Before writing SDK code
- Install the agent skill or read Agent skill
- Read Best practices — what a good receipt looks like
- Trust boundary:
@checkafter,@approvebefore init()— zero required args, never throws- Mark
estimated: trueuntil a check is a hard fact - Shipped vs backlog: Introduction status table