Etherscan for AI agents. Every OpenClaw run becomes a TEE-signed, encrypted receipt anchored on 0G. Share one URL — anyone verifies the agent ran exactly what it claimed. No wallet
AGENTSCAN is the Etherscan for AI agents.
The problem: AI agents are unauditable black boxes. When an agent claims it ran a tool, called
an API, or made a decision — there's no way for anyone else to prove what actually happened
inside the run.
AGENTSCAN fixes this with a primitive: every OpenClaw agent run is captured by our plugin,
each tool call is signed inside a TEE container, the full session log is encrypted client-side
and stored on 0G Storage, and an ERC-7857 receipt is minted on 0G Chain anchoring the root
hash. The reveal key stays with the owner — share one URL and anyone can verify, cold, that
the agent ran exactly what it claimed. No wallet, no login.
Built on the full 0G stack:
• 0G Storage — encrypted session logs
• 0G Compute — TeeML inference signed at source
• 0G Chain — Aristotle mainnet, AgenticID at 0xC6f7fB15...8937
• Agent ID (ERC-7857) — verifiable receipts as iNFTs
• Privacy / Secure Execution — encrypted-by-default, TEE-rooted
Live demo: https://mainnet.agentscan.online/verify/0 — a real DeFi-swap simulation receipt
anyone can verify in 30 seconds.
This is the audit trail layer for the agent economy that's about to land in 2026–2027.
built AGENTSCAN end-to-end in 11 days. OpenClaw plugin captures tool calls, TEE signs each
entry, encrypted log stored on 0G, ERC-7857 receipt minted on 0G Chain. Share a URL, anyone
verifies the agent ran exactly what it claimed.
Mainnet live on 0G Aristotle. Plugin on npm. Dashboard + docs deployed via Coolify. v0.3.0
added encrypted-by-default privacy with key-blind server
none