Cursor launches Auto-review: a classifier agent that governs coding agent autonomy by risk level

Cursor 推出 Auto-review 机制：用分类器智能体动态管控智能体自主权限

Cursor added Auto-review, a small classifier agent that checks tool calls before execution and decides whether to allow, block, or redirect them. Low-risk actions pass through; high-risk ones get blocked with feedback so the parent agent can try a safer approach without bothering the user. The classifier inspects files and workspace context instead of judging commands in isolation. The team found that a small model with some reasoning beats a pure speed model on both accuracy and latency. The post does not disclose exact latency numbers or classifier parameter count.

Why it matters: Cursor's first public write-up on agent safety architecture, with concrete model-selection tradeoffs useful to practitioners. The post doesn't disclose false-positive rates or user interruption frequency, so the score stays at 78 rather than higher.