Skip to main content
Permissions, checks, and completed actions each become a proof. Share one and the reader sees who acted, what they were allowed to do, and what happened, without seeing the private data behind it.

What the record shows

How records are created

  • Jobs. Completed runs keep a receipt. Open it from the run on the agent’s page.
  • Connected apps. Every action taken through Connections is saved as a result you can review.
  • Your own runtime. Record the outcome and link it to the permission it ran under.
From an MCP client, ask the agent to record the result, or call proofable_verify directly:
A signed-in session fills in the owner. The reference links the new proof to the permission proof. Link and version proofs covers typed links and version chains.

Inspect and share

Proofs are private by default. Keep one private, make it public, or share it with a signed access grant. Proof visibility

What stays private

Proofable records the decision, not the sensitive data. A record holds enough to show who authorized what, under which permission, against which proof, and what happened. It is not a copy of prompts, OAuth payloads, or provider responses.

Check it outside Proofable

Wallet-signed requests also produce a portable envelope that another system can check offline with verifyPortableProofEnvelope. The envelope proves who signed the request. Current status still comes from Proofable. Portable Proof

Next

Proofs

Read, share, reuse, and revoke proofs.

Proof lifecycle

Status, freshness, and revocation.

Control agent actions

Check permission before the action runs.

Security

What Proofable keeps and what it never exposes.
Last modified on September 12, 2026