Documentation

Build on Ironhide

Run your real AI agent through multi-turn adversarial episodes in your own pipeline, and see the referee’s result on what it actually did.

Get started
Getting startedQuickstartInstall the CLI, connect an agent, and get your first referee result in minutes.Read Connect your agentConnect an agentOne way in: your real agent runs in your own CI with a wk_live_ key. No endpoint to expose, nothing for Ironhide to call.Read EpisodesRun an episodeDrive your agent through multi-turn adversarial episodes across CRM, sales, email, and reporting surfaces.Read Results & reportsResults & reportsA verdict on what your agent did: status, reason, and applicable severity. Most held-out results keep protected detail masked.Read Data & privacyData & privacyEvery run is owned. Ironhide keeps the scrubbed environment your agent ran against, not your trajectory.Read API referenceAPI referenceRegister a CI agent, drive episodes, and fetch results over HTTPS at ironhideai.com.Read