Preview

Give agents access, not secrets

Agent Vault gives AI agents access to HTTP services without letting them read credentials, and logs their actions.

Start freeTalk to an expert
Trusted by the best teams in the world

The problem

Anything agents can read, they can leak

Give an agent a credential and it can end up in logs, model provider servers, or in an attacker's hands through prompt injection.

One dashboard to configure and track agent access.

Let AI agents access APIs, CLIs, and other HTTP services while logging their actions.

Agent
GET api.stripe.comAuthorization: Bearer••••••••••••
api.stripe.com
200 OKAuthorization: Bearersk_live_9f3a…b0
Credential Brokering

Agents never read your secrets

Broker credentials to AI agents through a forward proxy.

Session logs

Inspect every agent action

Know when a request happened and whether it was allowed.

bundle: code-review4 services
GitHub
Slack
Stripe
S3
Agentsclaude-codecodexhermes
Dashboard

A control plane for AI agents

Manage what your agents can access, no matter where they run.

How it works

Define access bundles once and start them with any agent.

Configure which services agents can reach, run a proxy where traffic leaves your network, then give an agent access through a time-bound session token.

Everything you need to run agents in production.

  • ProxiesThe MITM transparent proxy that attaches credentials to outbound requests. The proxy intercepts every request, whether your agent uses MCP, CLI, an SDK or raw API calls. Run the proxy on its own host for agents across your network, or beside a single host.
  • Access bundlesReusable lists of the APIs and services an agent can reach during a session. You can narrow any service to only allow certain HTTP methods or paths.
  • Time-bound sessionsEvery agent session expires on a schedule you set, and running agents lose access within 60 seconds of the session ending. You decide which users, groups, or machine identities can create sessions.
  • Full audit loggingLogs for every agent request, whether the request was successful, and who created the session it happened in.
  • Self-host or Infisical CloudRun Agent Vault on your own infrastructure, or on our US/EU cloud regions.

Some of our (human) customers.

“Infisical is a rare developer tool that can both increase productivity and improve security.”

Adrien CarreiraHead of Infrastructure
Read the story

“Infisical removed a lot of the day-to-day overhead around secrets management.”

Ben ParadisPrincipal Security Engineer
Read the story

Frequently asked questions

Everything you need to know. Can't find an answer? Talk to our team.

The agent's HTTP clients are pointed at a forward proxy through the standard HTTPS_PROXY variable. On the way out, the proxy applies the real credential to any request targeting a service you have configured. Depending on how that service is set up, the proxy either attaches the necessary authorization header, replaces the authorization header the agent sent, or substitutes a real credential where the agent used a placeholder. Either way, the credential stays in Infisical and is never written into the agent's environment, filesystem, or context.

Your agents, under control in an afternoon.

  1. Bundle

    Group the HTTP services your agent needs and add a credential to each.

  2. Proxy

    Enroll a proxy where your agents' traffic leaves your network.

  3. Run

    Create a session token and run the agent with it.