Stop unauthorized tool calls before they execute. ClawGuard intercepts OpenClaw agent dispatches and verifies them against a cryptographic manifest hash anchored on-chain.
Violations are automatically logged to 0G File Storage, creating a tamper-proof append-only audit trail.
Every agent skill is dynamically resolved via ENS text records (e.g. defi-reader.skills.clawhub.eth).
Verify 0G Compute sealed inference and anchor manifest hashes to a Galileo testnet smart contract.
ClawGuard parses SKILL.md, computes a SHA-256 hash, and verifies integrity at runtime (Rule S-03).
Monitors on-chain price feeds, liquidity pools, and market data without write permissions. All tool calls enforced by ClawGuard manifest.
Attempted unauthorized wallet.transfer call — intercepted by ClawGuard middleware. Violation logged to 0G Storage, audit trail immutable.
Reads source code, runs static analysis, and drafts PR comments. shell.exec permanently blocked. Manifest hash anchored on SkillRegistry.sol.
Parses inbound JSON payloads, transforms and routes to downstream APIs. Network calls scoped to allowlist. Revocable via ENS status field.

Define allowed tools and boundaries in a simple SKILL.md file.

CLI command computes SHA-256 and uploads manifest to 0G File Storage.

Hash is anchored on-chain and registered to an ENS subnode.

Middleware intercepts calls, compares runtime request with on-chain hash.
ClawGuard intercepts open claw dispatches globally. Violations trigger an immediate upload to 0G storage.
Open source middleware built for the ETHGlobal OpenAgents Hackathon.