Skip to content

IDProva launches April 7 — Registry packages coming at launch. Build from source now.

FAQ

IDProva is an open protocol for establishing verifiable identity, scoped delegation, and auditable action tracking for autonomous AI agents. It provides a standard way for agents to prove who they are, what they’re authorised to do, and produce tamper-evident records of their actions.

“ID” (identity) + “Prova” (Italian/Portuguese for proof, evidence, test). Pronounced “id-PRO-vuh.”

Yes. The core protocol and reference implementations are licensed under Apache 2.0. The protocol specification is freely available at idprova.dev.

IDProva is built by Tech Blaze Consulting, an Australian cybersecurity consultancy led by an ASD-endorsed IRAP Assessor with decades of experience in IT security and compliance assessment.

OAuth was designed for human-delegated application access. IDProva is purpose-built for autonomous AI agents with:

  • Agent-specific metadata (model, runtime, configuration attestation)
  • Delegation chains with scope narrowing (not just flat scopes)
  • Hash-chained audit trails (not just access logs)
  • Progressive trust levels (not binary authenticated/unauthenticated)
  • Post-quantum cryptography from day one

SPIFFE provides workload identity — identifying what is running. IDProva provides agent identity — identifying who an agent is, what it’s authorised to do, on whose behalf, and what it has done. IDProva adds delegation semantics, audit trails, and compliance mapping that SPIFFE doesn’t address.

AI agents may have long-lived identities (months to years). Adversaries can “harvest now, decrypt later” — capturing signed tokens today for future quantum cryptanalysis. IDProva’s hybrid Ed25519 + ML-DSA-65 approach provides protection against both classical and quantum attacks from day one.

IDProva provides binding specifications for:

  • MCP (Model Context Protocol) — Agent tool calls and resource access
  • A2A (Agent-to-Agent) — Inter-agent communication
  • HTTP — Standard API authentication

IDProva layers on top of these protocols — it provides identity and delegation; the underlying protocol provides transport.

No. IDProva supports multiple resolution methods:

  1. Well-known endpoints — Publish DID Documents at /.well-known/did/idprova/
  2. Self-hosted registry — Run your own registry server
  3. Managed registries — Use a hosted registry service
  4. Universal resolvers — Standard DID resolution infrastructure
  • BLAKE3 — Primary hash algorithm (fast, secure)
  • SHA-256 — Interoperability fallback
  • Ed25519 (required) — Classical signatures
  • ML-DSA-65 (recommended) — Post-quantum signatures (FIPS 204)

Does IDProva help with NIST 800-53 compliance?

Section titled “Does IDProva help with NIST 800-53 compliance?”

Yes. Action Receipts directly satisfy multiple NIST 800-53 controls:

  • AU-2 (Auditable Events), AU-3 (Content), AU-8 (Timestamps), AU-9 (Protection), AU-10 (Non-repudiation)
  • IA-2 (Identification & Authentication), AC-6 (Least Privilege)

Does IDProva help with Australian ISM compliance?

Section titled “Does IDProva help with Australian ISM compliance?”

Yes. The protocol maps to ISM controls including ISM-0585, ISM-0988, ISM-0580, and ISM-1405.

Yes. Action Receipts map to SOC 2 Trust Services Criteria CC6.1, CC6.2, CC6.3, and CC7.2.

Has IDProva been submitted to any standards bodies?

Section titled “Has IDProva been submitted to any standards bodies?”

IDProva was submitted to the NIST Center for AI Safety and Identity (CAISI) via RFI NIST-2025-0035 on Security Considerations for AI Agents. We are also engaging with the NCCoE AI Agent Identity & Authorization concept paper process.

  • Rust — Full implementation (idprova-core, idprova-cli, idprova-registry)
  • Python — SDK in development
  • TypeScript — SDK in development
Terminal window
cargo install idprova-cli

See the Quick Start guide — you’ll have an agent identity in under 5 minutes.