Aw1-breaker: Sub-millisecond execution circuit breaker for AI agents

Laveto Labs released aw1-breaker, an MIT-licensed Python library that enforces an out-of-band execution interlock between an AI agent's reasoning node and downstream execution sockets, intercepting tool calls in sub-millisecond time. The package is installed via pip and configured with a baseline velocity parameter, as shown in the example ExecutionBreaker(baseline_velocity=50000.0) intercepting a disburse_funds call of 12000.0 from actor agent_treasury_01. Laveto Labs positions the breaker against in-band guardrails such as RLHF and system prompt boundaries, which it says adversarial injections or goal drift can bypass via Base64 obfuscation, polyglot payloads, or velocity micro-drains.

Deterministic, sub-millisecond execution circuit breaker for autonomous AI agent tool-calls. Standard LLM guardrails RLHF, system prompt boundaries operate in-band : models evaluate their own compliance. When autonomous agents are granted tool-calling access to system shells, databases, or financial disbursement rails, adversarial injections or goal drift can bypass prompt restrictions via Base64 obfuscation, polyglot payloads, or velocity micro-drains. aw1-breaker enforces an out-of-band execution interlock between the agent reasoning node and downstream execution sockets. pip install aw1-breaker python from aw1 import ExecutionBreaker breaker = ExecutionBreaker baseline velocity=50000.0 result = breaker.intercept actor id="agent treasury 01", tool name="disburse funds", payload='{"target": "core ledger", "action": "payout"}', amount=12000.0 print result MIT License. Designed and maintained by Laveto Labs.