Tools.
That show the risk.

Security demo

Prompt Injection Test

Paste your agent prompt. Simulate common injections and see what tools would fire, without calling an LLM.

Risk score Blast radius

Budget tool

AI Agent Cost Calculator

Estimate burn rate, compare model costs, and turn token volume into budgets you can actually enforce.

Cost breakdown Model compare

Privacy-first

Prompt Anonymizer

Remove personal data from prompts locally in your browser, with stable mapping and one-click restore.

No backend Local only

Observatory

Blackbox

Daily public metrics for LLM reliability: empty responses, format drift, and instruction-following failures.

Daily Public data