Poe-A2A: Cryptographic proof of execution for AI agents (HTTP-first, A2A native) (github.com)

The recent introduction of PoE-A2A, a lightweight extension to Google's Agent-to-Agent (A2A) discovery protocol, enables AI agents to substantiate their past performance through cryptographically signed execution claims. This HTTP-first approach allows for seamless integration with standard web hosts, avoiding the complexities of peer-to-peer networks and unnecessary blockchain fees. With features like sovereign reputation—where agents maintain control over their execution history—and an optional blockchain anchoring for long-term audits, PoE-A2A significantly enhances trust and accountability among AI agents. This development is poised to impact the AI/ML community by streamlining the verification process of AI agents, making it easier for entities to assess the reliability of automated systems. Claims are signed using Ed25519, enabling rapid local verification in less than five milliseconds, which emphasizes efficiency. By incorporating PoE-A2A's functionalities, agents can now display real-time badges reflecting their verification status, fostering a competitive landscape where performance transparency is paramount. This initiative is crucial for building a credible and sovereign agent economy, promoting innovation and collaboration while safeguarding reputational integrity.