Image 1 of 1
Image 1 of 1
Image 1 of 1
Image 1 of 1
Decreed proves what your AI agents were allowed to do.
When a policy decides whether an AI agent may act, Decreed records the decision as a signed authorization receipt: the action that was asked for, the agent and the person behind it, the delegation chain, the policy and the engine that decided, the decision, and the evidence it cited. Every signature says which cryptographic module made it.
How it works: 01 Decide — your policy engine allows, denies or holds the action, as it does today. 02 Receipt — Decreed signs a receipt of the decision. 03 Log — the receipt goes into a transparency log, with a proof of where it sits, so any later change shows up when it’s checked. 04 Countersign — the receiving system signs that it ran the action, or that it didn’t. 05 Verify — an auditor checks the receipt offline, with an open-source verifier and the keys they choose to trust.
What verification proves, and what it doesn’t: a receipt that checks out hasn’t changed since it was signed, was signed by a key you trust, and sits where it says in the log, with its delegation, scope, approval and timing rules holding as stated. It doesn’t prove that the policy’s decision was right, that the evidence it cites is true, or that every agent action was recorded. Decreed works with Open Policy Agent, MCP gateways, OpenTelemetry and agent frameworks.
Request a technical walkthrough or a design-partner discussion at cognitoagora.com.
Decreed proves what your AI agents were allowed to do.
When a policy decides whether an AI agent may act, Decreed records the decision as a signed authorization receipt: the action that was asked for, the agent and the person behind it, the delegation chain, the policy and the engine that decided, the decision, and the evidence it cited. Every signature says which cryptographic module made it.
How it works: 01 Decide — your policy engine allows, denies or holds the action, as it does today. 02 Receipt — Decreed signs a receipt of the decision. 03 Log — the receipt goes into a transparency log, with a proof of where it sits, so any later change shows up when it’s checked. 04 Countersign — the receiving system signs that it ran the action, or that it didn’t. 05 Verify — an auditor checks the receipt offline, with an open-source verifier and the keys they choose to trust.
What verification proves, and what it doesn’t: a receipt that checks out hasn’t changed since it was signed, was signed by a key you trust, and sits where it says in the log, with its delegation, scope, approval and timing rules holding as stated. It doesn’t prove that the policy’s decision was right, that the evidence it cites is true, or that every agent action was recorded. Decreed works with Open Policy Agent, MCP gateways, OpenTelemetry and agent frameworks.
Request a technical walkthrough or a design-partner discussion at cognitoagora.com.