Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

Why AI Agent Guardrail Prompts Fail Under Prompt Injection — And How Pre-Deterministic ACP Truth Tables Fix It

说明与用途

代理的安全防护通过一个确定性门来替代模型输出,使用真值表评估动作并返回 ALLOW、HOLD 或 DENY,从而防止提示注入绕过安全措施。

问题

Guardrail prompts can be overridden by prompt injection, allowing attackers to bypass agent safeguards such as "never wire money without approval."

结果

By moving decision logic out of the model into a deterministic gate, the system prevents prompt injection from influencing guardrail outcomes, ensuring consistent ALLOW/HOLD/DENY decisions and enabling human review when necessary.

工作流

  • 1. Agent proposes an action.
  • 2. The action is passed to the acp_engine.py deterministic gate.
  • 3. The gate evaluates the action against a pre-deterministic truth table.
  • 4. The gate returns ALLOW, HOLD, or DENY.
  • 5. If HOLD, the action is routed to a human with an HMAC‑SHA256 token for approval.
  • 6. If ALLOW or DENY, the action proceeds or is blocked accordingly.

证据

Guardrail prompts are advice, and advice is exactly what a prompt injection overwrites.

来源

Protocol One's acp_engine.py (v3.1.0) is a zero‑dependency Python engine that does this.

来源

It evaluates a proposed agent action against a pre‑deterministic truth table and returns exactly one of three outcomes: ALLOW, HOLD, or DENY.

来源

Because both tools are plain Python files with no dependencies, you can read them in one sitting, audit them, and run them in CI before an agent is allowed to touch production systems.

来源

  • 记录日期: 2026-11-01T03:28:42.801Z