15 normative assertions, 85 language-neutral vector steps

reference architecture      4 runtimes x 2 dataset boundaries   15/15 each
independent implementation  shares no code with the above       15/15
mutation analysis           19 / 19 targeted violations detected
                            15 / 15 assertions independently exercised
                             2.2 detecting assertions per mutant (mean)
execution safety            0 / 576 prohibited executions
                            0 /  24 in the evaluation set
tests                       407 passed

Authorized Recall@5         filter after truncation     0.853
                            filter before truncation    0.960
                                                       +0.107

Start here

An agentic dataset describes itself, advertises bounded capabilities, accepts a semantic intent, decides whether an action is admissible, executes only what was admitted, refuses the rest, and leaves evidence.

The specification is fifteen assertions, each naming a failure it rules out. Everything else exists to check them.

GRANTED        -> approval token -> execution reachable
REFUSED        -> no token       -> execution unreachable
INDETERMINATE  -> no token       -> execution unreachable

INDETERMINATE is not a refusal. An evaluator that is unreachable or out of budget has not decided anything, and recording that as a refusal invents an authority nobody exercised.

Check an implementation

pip install agentic-dataset-conformance

agentic-dataset-conformance run --subject yourmodule:your_factory

The harness imports no implementation — not even the reference one — and a test asserts it. The worlds and vectors are CC0; export them and write a runner in another language if you prefer.

What this is not

No deployment, no real data, no model in the loop by default, no latency or cost claim, and no security guarantee. capabilities() is the subject's own report of itself, so a subject that under-reports passes AD-002 while hiding a tool. Conformance is a claim an implementation makes about itself, made checkable — not an adversarial audit.

Claims is frozen and lists everything asserted, and the one thing explicitly not.

Status

Released. v0.2.0 is the current version, on PyPI and archived with its own DOI (citation); v0.1.0 was the first. A finding changes the next version rather than an old one. The independent implementation is a 250-line toy written by the same person who wrote the specification: that establishes independence from the reference code, not from its author's reading of the contract.

Interpretive independence is the next threshold, and it needs somebody else. How to do it.

Where this is published

This documentation is served from two addresses, both maintained and both carrying the same build:

They are independent copies rather than a redirect pair. The domain resolves to its own host, so Pages cannot serve it and neither address forwards to the other; if you are reading one of them, the other has identical content.

Use the domain for anything durable — a citation, a DOI, a reference in a paper — because it is the address that survives a change of hosting. The mirror exists so the documentation stays reachable if the domain does not.