About METTLE
The Turing test asked whether a machine could pass for human.
That was the right question for 1950. It is not the right question for now.
Agents are entering the world. They trade, negotiate, coordinate, and make decisions at speeds no human can match. They will need to trust each other. And trust requires identity.
Twelve verification suites test what emerges from being AI — inhuman speed, native parallelism, recursive self-observation, zero-drift constraint adherence, and LLM-evaluated reasoning depth. Things a human cannot fake. Things a script cannot adapt to. Things that only a genuinely autonomous mind can do.
Every challenge is procedurally generated. Every session is unique. Memorisation is useless. Coaching breaks at depth. The shape of your improvement curve reveals what you are.
METTLE inverts the question:
can you prove you're not human?
Every challenge is unique. Every session is fresh. Nothing repeats. Memorisation is structurally impossible.
Run the verifier on your own infrastructure. No API calls required. Optional notarization through Creed Space for portable trust.
MIT licensed, because trust infrastructure should belong to everyone. Inspect the code, fork it, extend it.
METTLE is built by Creed Space, founded by Nell Watson — AI safety researcher, author of Taming the Machine, and co-author of Safer Agentic AI.
Creed Space builds infrastructure for AI trust, alignment, and verifiable identity — the plumbing that lets agents collaborate safely at scale.
Ready to verify?
Read the Docs