Integrations: Policy engines

Argent decides with Cerbos

If your authorization rules already live in Cerbos, Argent can ask your Cerbos PDP about every AI agent request. Each agent's permissions arrive as Cerbos roles, and only an explicit allow lets the request through.

Cerbos logo
01 / INTEGRATIONS: POLICY ENGINES

What Argent does with Cerbos

Keep the policy engine your team already runs. Argent asks it before any agent request goes out.

NIST's zero trust architecture, SP 800-207, puts it plainly: "Authentication and authorization (both subject and device) are discrete functions performed before a session to an enterprise resource is established." (NIST SP 800-207). Argent applies that to AI agents on every request: a policy engine decides, not the agent.

Most tenants use Argent's built-in OPA engine. A tenant that already writes its authorization rules in its own engine can route Argent's decision there instead, so one set of rules covers people, apps and AI agents.

  • Agents become Cerbos principals. The agent is the principal, and the permissions it holds in Argent become its Cerbos roles. The rest of the request rides as attributes your policy can read.
  • One resource kind per kind of traffic. By default the resource kind is the request class, llm, mcp, a2a, or unknown for anything else, so you can write a separate policy for model calls and tool calls without extra setup.
  • Only an explicit allow. The request passes only on EFFECT_ALLOW. EFFECT_DENY, no matching policy, or no effect at all is a deny.
  • Validated end to end. This connector is live-validated: a real allow, a real deny and an unreachable PDP that fails closed.
02 / INTEGRATIONS: POLICY ENGINES

How it connects

Argent calls Cerbos's own Check API. Your Cerbos PDP stays as it is.

DirectionArgent calls your engine once per request, at the endpoint you give
CallPOST <endpoint>/api/check/resources, one principal against one resource
Sign-inNone (a network-isolated PDP), an API key sent as a bearer token (for a PDP behind a proxy that checks it), or mutual TLS
CertificatesCertificate checks stay on. A self-signed engine is trusted by pinning its CA certificate, never by turning verification off
NetworkArgent refuses private, loopback, link-local and cloud metadata addresses unless an operator has allowed that network range; the check runs when you save, with a reason
CredentialStored as a reference in Argent's secret store, never in the connector settings, and dropped when the decision is done

Cerbos documents CheckResources as "the main API entrypoint for checking permissions for a set of resources" (Cerbos API).

03 / INTEGRATIONS: POLICY ENGINES

What your policy sees

Your engine gets the same facts on every request. None of them is a secret.

FieldWhat it holds
identityThe agent making the request, as Argent resolved it from its key
scopesThe permissions that agent holds in Argent, sent as its Cerbos roles. An agent with none is denied.
tenantThe tenant the request belongs to
request_classThe kind of traffic: llm, mcp, a2a, or unknown for anything else
resourceThe request path
destinationThe upstream host the request is bound for
phaseheaders: your engine decides before the request body is read
modelEmpty at the headers phase, because the body is not read yet

No prompt, no request body, no key and no credential is ever sent to your engine. Because your engine decides before the body is read, Argent's own model check still runs on the body afterwards, so a model nobody approved is refused as well.

04 / INTEGRATIONS: POLICY ENGINES

What you configure

One connector, in the console. Your Cerbos policies decide.

  1. Write the policy. Add a resource policy for the kinds llm, mcp, a2a and unknown, or one kind of your choosing, with an access action allowed for the roles you trust.
  2. Open Policy Engines. In Integrations, open Policy Engines and choose Cerbos.
  3. Enter the PDP. The endpoint, such as https://cerbos.internal:3592.
  4. Name the action and kind if you need to. The action defaults to access; the kind defaults to the request class. Override either to match an existing policy.
  5. Pick the sign-in. None, API key or mutual TLS, with the reference to the stored credential. Optionally pin your CA certificate and set a time limit up to 30 seconds.

Test the connection from the same page. Once the connector is active, every request from that tenant's agents is decided by your engine. Connecting an engine is an administrator's job: the Integrations page is gated by console role, like every console page.

05 / INTEGRATIONS: POLICY ENGINES

Fail closed, every time

A policy engine Argent meant to ask but could not reach never allows anything.

  • Timeout is a deny. Each decision has a time limit, 3 seconds unless you set another, and never more than 30 seconds. No answer in time means the request is refused.
  • Errors are denies. A network error, a status other than success, or a reply Argent cannot read is a deny.
  • No quiet fallback. Argent never falls back to its built-in engine when yours fails. Your engine may be stricter than the default, so substituting the default would be a way around your policy.
  • Named reasons. A refusal from your engine is recorded as pdp.external.deny, an engine that failed to answer as pdp.external.error, and one Argent could not build or look up as pdp.external.unavailable. The caller gets a 403; your engine's reply body and headers are never passed back to it.
  • Signed record. Every refusal goes on Argent's signed record with its reason, so an auditor can see which engine said no and why. See One-Click Audit Evidence.
  • One tenant, one engine. The engine is chosen per tenant from the authenticated request, never from anything the caller sends. A tenant with no engine connected uses Argent's built-in OPA engine.
Other policy engines

Argent decides with its built-in OPA engine or with the engine your team already runs, and filters retrieval with OpenFGA.

  • Open Policy Agent (built in)
  • AuthZed SpiceDB
  • Topaz
  • OpenFGA for RAG