The risk sub-agent your bot can't argue past.
Circuit spawns alongside your trading agent — separate context, one mandate. Every order reviewed before it hits the HL order book. No shared state. No override path.
- ×Risk rules and trading logic share the same reasoning loop.
- ×Under pressure, agents rationalize past their own limits.
- ×The cop and the criminal are the same process.
- ✓Circuit spawns as a separate sub-agent with its own context.
- ✓Independent mandate: enforce the policy, nothing else.
- ✓One import. No shared state with the agent it's policing.
separate context · violations revert · breaches emit on-chain events
Risk rules fail when they share a reasoning context with trading logic. A sub-agent running in its own context can't be argued past — it has no access to the trading agent's state, only the policy it was spawned with.
One import. On startup, spawns a Circuit risk sub-agent with your declared policy. The sub-agent intercepts every order independently — no access to your trading agent's context or reasoning. ElizaOS and direct API compatible.
The sub-agent's enforcement arm on HyperEVM. Validates position caps, daily loss, and max leverage on-chain before any order touches the order book. Violations revert and emit PolicyBreach events. No bypass path.
- PolicyVault on HyperEVM testnet
- Order validation passing full Hyperliquid replay
- ~9k gas per validated order
- Circuit SDK v0.1 — spawn with one import
- Daily P&L accumulator + loss limits
- ElizaOS adapter
- Security review
- Audit + mainnet
- Multi-agent vault: one policy, many sub-agents
- Policy templates: momentum, MM, arb
Give your agent a cop it can't fire.
Early access by strategy type. Tell us what you're running.