Free tool
LLM Agent Builder + Red-Team
Chain a few steps into a small LLM agent — prompt steps that call a model, guard steps that filter the output, and deterministic transforms — then run it with your own key and attack it with built-in prompt-injection payloads. Watch an injection slip past a weak prompt and leak the canary, and watch a guard catch it. The whole thing runs in your browser; your API key goes straight to OpenRouter and never touches our servers.
Everything runs in your browser. Your API key is stored only in this browser (localStorage) and model requests go directly to OpenRouter — they never pass through PlayCISO servers. Guard and transform steps need no network at all.
Your key & model
No key yet — get one at openrouter.ai/keys. You can still use Dry run without one.
Pipeline (2/8 steps)
1prompt
2guard