Docs/Cookbook/Agent Guard CookbookPublishedGuidanceLAST REVIEWED · 30 AUG 2026Agent Guard CookbookBind agent action to identity, policy, capabilities, and approval.Agent Guard recipesProtect an MCP tool→Issue Agent Identity→Bind a mission→Create scoped capability→Require human approval above risk threshold→Revoke an agent→Protect an API with Agent Guard Edge→End of guideWas this page helpful?YesNot yetPrevious← FinOps CookbookNextEverywhere Cookbook →