ATP is an open, transport-agnostic protocol that lets AI agents produce cryptographically verifiable proofs of task execution. It adds an accountability layer to multi-agent systems — enabling trust assessment, reputation scoring, and compliance verification without exposing sensitive task content.
Full specification and reference: agenttrustprotocol.org
See CONTRIBUTING for how to propose changes, open issues, or submit PRs. Bug reports and clarifications go to the GitHub Issues tracker.
For direct inquiries: atp@aiquilibria.com
Apache 2.0 — see LICENSE.