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

# Run AI agents securely

> Give AI agents access to the services they need without ever handing them a real credential.

Infisical Agent Vault is an access layer for AI agents. Agents route their API calls through a proxy that attaches the real credentials on the way out, so the agent itself never holds them. If a [prompt injection](https://en.wikipedia.org/wiki/Prompt_injection) tricks the agent into revealing a credential, there's nothing there to reveal.

<Warning>
  Agent Vault is in preview and subject to change. We don't currently recommend it for production use.
</Warning>

<Card title="Agent Vault docs" icon="vault" href="/docs/documentation/platform/agent-vault/overview">
  Access bundles, services, sessions, proxies, and access control.
</Card>

## Why brokering access matters for agents

Traditional secrets management involves returning credentials back to applications and services. This isn't suitable for AI agents because a credential in the agent's environment can leak in many ways. An agent may inadvertently include it in a log line, a message, or a tool call it makes, or an attacker may deliberately extract it through [prompt injection](https://en.wikipedia.org/wiki/Prompt_injection).

Infisical Agent Vault solves this by brokering credentials using a proxy at the network boundary. The proxy attaches real credentials to services you've allowed, and only until the session expires (or you revoke it).