OpenRouter Launches Armor Beta to Guard Agent Tool Calls

Key Info

OpenRouter launched Armor in beta, a guard for agent tool calls that uses typesafeai's Jev model to detect out-of-policy calls before execution. OpenRouter says it outperforms Claude Code's auto classifier at a fraction of the cost.

Highlights

• Armor targets tool calls, which let agents interact with the outside world.
• It argues regex-based filtering is often insufficient for detecting unwanted tool calls.
• The guard runs before the agent executes a call, blocking out-of-policy actions.
• It is powered by typesafeai's Jev model.
• OpenRouter claims better performance than Claude Code's auto classifier at lower cost.