One protocol.
Every agent.
The Elydora Responsibility Protocol (ERP) defines the open standard for binding AI Agent actions to verifiable, tamper-evident, legally defensible responsibility records. ERP is to agent accountability what OAuth is to web identity.
Four core primitives.
Signed operation envelope submitted by an Agent. Contains agent_id, operation_type, payload_hash, nonce, timestamp, Ed25519 signature, and prev_chain_hash for sequential chaining.
Cryptographic chain linkage hash per Agent. SHA-256 hash of the canonicalized operation record that links each operation to its predecessor, creating an immutable per-agent chain.
Platform-signed admission receipt returned for every accepted operation. Contains receipt_id, operation_id, chain_hash, seq_no, and the Elydora platform signature.
Merkle root anchoring a batch of operations into a single verifiable hash. Follows RFC 9162 transparency-style trees. Optionally anchored via RFC 3161 TSA.
Data flow.
Built on open standards.
Digital signatures for operation records
Hash function for chain linking and content integrity
JSON Canonicalization for deterministic serialization
Certificate Transparency v2 style audit proofs
Trusted timestamping for epoch root anchoring
Key lifecycle and rotation guidance
Build on the protocol.
Integrate the Elydora Responsibility Protocol into your agent infrastructure. Read the developer documentation or get in touch.