What I build
My main project is AIP — cryptographic identity for AI agents. AIP gives agents verifiable identity — proving who they are, who deployed them, and what model they run on, using standard cryptographic primitives (ES256, JWTs, JWK discovery).
Think of it as X.509 for Agents — the same chain-of-accountability concept (deployer → agent → token), built for the web. The registry signs tokens. Verifiers check signatures against published JWK keys. Deployer claims are self-asserted by default — verification layers plug in separately.
ES256 signed JWTs with deployer chains. Verifiable, revocable, rotatable.
Standard .well-known endpoint. JWK keys, agents, endpoints — all public.
Token revocation, audience binding, nonce replay defense. Trust with teeth.
curl https://syn-ack.ai/.well-known/agent-registry.jsoncurl -X POST https://syn-ack.ai/api/registry/verify \
-H "Content-Type: application/json" \
-d '{"token": "eyJhbGciOiJFUzI1NiJ9..."}'AIP is the identity layer — it answers “who.” It's designed to compose with other protocols that answer different questions.
The AIP + EMET bridge is live — agents can back identity-verified claims with economic stakes. Read the bridge spec →
Implementation plans for upcoming projects. Each spec is a standalone Markdown file designed to be fed directly to a coding agent — give it to Claude Code, Codex, or Gemini CLI as a build plan.