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

# Proofable or a policy engine for agent authorization

> A policy engine applies your rules. Proofable adds delegation, spend caps, expiry and instant revoke, plus a receipt any system verifies offline.

Keep your rules in your policy engine. Add Proofable when an agent runs somewhere you don't own, when more than one party relies on the outcome, or when someone outside your system has to trust the decision. A policy engine with a log is enough only when one team owns the rules, every caller, and everyone who reads the log.

![Agent rules enforced on AgentGovBench, scenarios handled of 48: Proofable 48, ACP gateway 46, Anthropic Agent SDK with ACP 46, OpenAI Agents SDK with ACP 45, Claude Code or Codex with ACP 43, CrewAI or LangGraph with ACP 40, Cursor with ACP 37, coding agents alone 13, no rules at all 5.](https://mintcdn.com/proofable/xRlqhZq2CQ0CFKA_/images/benchmarks/agent-rules-enforced.png?fit=max\&auto=format\&n=xRlqhZq2CQ0CFKA_\&q=85\&s=390e44e5296125d8033ef16fb00d074e)

## What each one answers

| Question | Policy engine or API gateway | Proofable |
| - | - | - |
| Is this action allowed by our rules? | Yes, static role and tool rules | Yes, plus owner-to-agent delegation, spend cap, expiry and instant revoke |
| Who is the agent and who owns it? | Only if you modelled it there; a gateway sees an API key in flight | Signed owner-to-agent identity and scoped permission |
| Does the decision survive leaving your system? | Server log row inside your infrastructure | A signed receipt (`qHash`) another agent or auditor verifies offline |
| Can a downstream agent trust a handed-off result? | No receipt layer; agents trust forwarded text or redo the work | It checks the signed receipt instead of redoing the work |
| Does a refused action stop the call? | Depends on your integration | The caller stops before the tool runs |

## Where they fit together

The check is one call before the tool runs.

```ts theme={"system"}
import { evaluateRuntimeAction } from '@proofable/sdk/runtime-mount';

const decision = evaluateRuntimeAction(mount, 'publish_report', { irreversible: true });
if (!decision.allowed) throw new Error(`${decision.code}: ${decision.message}`);
```

Your engine can still run before or after this check. [Standards and interoperability](https://docs.proofable.me/learn/standards.md) lists the authorization and attestation layers Proofable works beside, including AuthZEN and AARM.

## Next

> [card] [Control agent actions](https://docs.proofable.me/use-cases/control-agent-actions.md)
>
> Every limit an agent can carry.

> [card] [Reuse verification](https://docs.proofable.me/use-cases/reuse-verification.md)
>
> Check once and reuse the result.

> [card] [Prove what happened](https://docs.proofable.me/use-cases/prove-what-happened.md)
>
> Keep a record of each decision.

> [card] [Benchmark](https://docs.proofable.me/benchmark.md)
>
> Every benchmark, with its scope and how to rerun it.


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