A zero-trust logic judges a claim written as a ZFL table. A compound claim is judged T or F; a claim that is a single name answers with that name's own value, which is Z while it is not verified. No model runs here — the table is yours, or your agent's.
Pick an example or paste a ZFL document, then press Judge.
| Field | Meaning | Values |
|---|
MCP endpoint — tools language, examples, judge. Your agent's own AI writes the table; ZTL judges it. Two more checkers, no AI: check_steps (does each step of an algebra derivation keep exactly the same solutions?) and judge_naive (sets defined by a property, self-reference allowed: what the definitions decide, and a passport — PARADOX, UNDERDETERMINED, … — for what they leave undecided), with checker_examples; in a browser at studio.ztljudge.com/tools.