OpenRouter launches Armor beta to screen agent tool calls before they run
Article timeline
The articles in this story. Times are ET.
OpenRouter@OpenRouterOfficialOpenRouter lets agents opt in to Armor via a plugin settingAIOpenRouter says any agent doing inference through its platform can enable Armor by adding it as a plugin with custom rules, using the configuration [{ id: "armor" }]. Documentation for the feature is available at
- OpenRouter@OpenRouterOfficialArmor scores agent tool calls against rules and user intent
AIOpenRouter says Armor checks each tool call against a set of rules and a slice of the transcript, prioritizing user intent. In its example, a git push is allowed when the user asked to push their work, while a dangerous command the user never requested is flagged.
- OpenRouter@OpenRouterOfficialArmor makes Claude Code's auto permissions mode work with any model
AIArmor says it makes Claude Code's auto permissions mode compatible with any model. OpenRouter's Ori command runs Claude Code with a chosen model, such as moonshotai/kimi-k3, and the --permission-mode auto flag.
- OpenRouter@OpenRouterOfficialOpenRouter launches Armor beta to screen agent tool calls before execution
AIOpenRouter launches Armor in beta, a guard that checks agent tool calls against policy before they run. Armor uses typesafeai's Jev model to catch out-of-policy calls that regex filtering often misses. OpenRouter says it performs better than Claude Code's auto classifier at a fraction of the cost.
