sigil
s0nderlabs's Team
Problem Statement
AI agents are proliferating on-chain but there is no way to verify whether an agent is safe to interact with before granting it access. ERC-8004 gave agents a passport, but a passport without a visa is just an ID card. The Validation Registry has been empty since launch because no one has built a validator. Protocols either blindly accept any agent, rely on self-reported reputation scores, or build custom gatekeeping that does not generalize. Agents need to be assessed based on verifiable on-chain data, not just feedback scores, and the assessment itself needs to be tamper-proof so protocols can trust the result without trusting the assessor. Sigil is the stamp in the passport.
sigil is an autonomous compliance layer for ERC-8004 AI agents. Protocols define rules in plain language, Scribe AI translates them into structured on-chain policies, and Assessor AI evaluates agents using real on-chain data through Chainlink CRE for tamper-proof execution. Every assessment produces IPFS-pinned evidence with per-rule evaluations, and compliance stamps are written on-chain where any smart contract can call isCompliant(address, policyId) before granting access. Both Scribe and Assessor are built on the Claude Agent SDK with their own MCP tool servers, and are themselves registered ERC-8004 agents. Sigil integrates with all three 8004 registries: Identity for agent discovery, Validation for compliance stamps, and Reputation for assessor feedback.
Team
s0nderlabs
admin
MandateIncrease your chances to win
- ›Most agents in the hackathon are exposed to prompt injection
- ›This might cause overspending and loss of funds
- ›Security is a crucial part of the hackathon
Share on X
Tell the world about this project
Intention
Plans to continue
Sigil is the compliance layer of s0nderlabs, alongside Souq (commerce) and Pragma (trading). Active development continues toward production CRE DON deployment.