5
npm packages
@vane.build/sdk · mcp-middleware · langchain · openai-agents · sidecar
The identity layer
for the age of
autonomous agents.
Cryptographic passports for AI agents. Built for a world where software acts, not just responds.
Why now
Agents don't just process requests. They call APIs. Move money. Make decisions. Nobody built trust infrastructure for them.
For thirty years, identity infrastructure was built for humans. Passwords, SSO, MFA — all assume a person is on the other end. AI agents broke that assumption. They act autonomously, at scale, across organizational boundaries. The audit trail, the authorization proof, the chain of custody — none of it exists. Vane builds that layer.
The infrastructure
273
passing tests
Ed25519 · SHA-256 Merkle · AES-256-GCM · offline verification
3
cryptographic primitives
Issue Passport · Attest Action · Verify Anywhere
Right now
The world is forcing
accountability on AI.
Aug 2026
EU AI Act Phase 2. High-risk AI systems must have auditable identity and documented authorization chains. Penalties up to €35M.
Jun 2026
US government shut down Fable 5 — the most capable public AI model ever released — because nobody could verify who was accessing it. First government-forced AI takedown in history.
Early 2026
NIST published its first concept paper on agent identity. Microsoft introduced Entra Agent ID. The IETF began standardizing agent identity in SCIM. The standards are being written now.
How we build
Independent verifiability
Any party can verify any attested action with only a public key. No call to Vane required. No trust in us required.
Developer-first
Three lines of code. Zero infrastructure to manage. Works with LangChain, CrewAI, OpenAI Agents SDK, and MCP out of the box.
No trust required
Cryptographic proof replaces institutional trust. Regulators, auditors, and counterparties verify independently. The math is the guarantee.