> ## Documentation Index
> Fetch the complete documentation index at: https://praison.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Egress Guard Protocol • AI Agent SDK

> EgressGuardProtocol: Contract for the wrapper's secret-egress firewall.

# EgressGuardProtocol

> Defined in the [**secrets**](../modules/secrets) module.

<Badge color="blue">AI Agent</Badge>

Contract for the wrapper's secret-egress firewall.

The real guard lives in the wrapper (`praisonai`): it inspects an
outbound request, decides via the operator host allowlist whether a
credential may leave, and either substitutes the sentinel for the real
secret (:func:`desentinelize`) or refuses with an audited event. Core owns
only this seam — no proxy, no host enforcement, no heavy imports — so a
deployment can inject a guard without pulling anything heavy into core.

## Methods

<CardGroup cols={2}>
  <Card title="sentinel_for()" icon="function" href="../functions/EgressGuardProtocol-sentinel_for">
    Instance method.
  </Card>

  <Card title="allow_egress()" icon="function" href="../functions/EgressGuardProtocol-allow_egress">
    Instance method.
  </Card>
</CardGroup>

## Source

<Card title="View on GitHub" icon="github" href="https://github.com/MervinPraison/PraisonAI/blob/main/src/praisonai-agents/praisonaiagents/secrets.py#L485">
  `praisonaiagents/secrets.py` at line 485
</Card>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.