An agentic dataset carries a machine-readable descriptor of itself, advertises a bounded set of capabilities, declares prohibitions, and is reached through a control plane that decides whether a requested action is admissible before that action runs.
The unit of governance is the decision to act, not the query. A conventional dataset answers or fails to answer. An agentic dataset additionally produces a verdict — approved, refused, or indeterminate — and an evidence record naming the descriptor clause that produced it.
| Repository | Contents |
|---|---|
reference |
The 15-assertion behavioural contract, its language-neutral executable vectors, a reference control plane, the conformance suite, Authorized Recall@K, and all measurements. |
programme |
What agentic datasets are, and where the licensing boundary sits. Prose only. |
showcase |
Five demonstrations of the contract, running the normative vectors client-side against a TypeScript subject. |
agentic-datasets.github.io |
The organization site. |
.github |
This profile, and the record of what the organization's settings say and why. |
Documentation: agenticdatasets.org/reference Showcases: agenticdatasets.org/showcase
15 normative assertions as 85 language-neutral vector steps · 15/15 portable through the public interface alone · 4 agent runtimes × 2 dataset boundaries, 15/15 each · an implementation sharing no code with the reference, 15/15 · 17/17 targeted violations detected by their named assertion · 0 prohibited executions across the measured matrix.
Every figure is reproducible from the reference repository and carried in its frozen claims table with its own caveat.
No security guarantee. A subject's capability report is its own account of itself, and one that under-reports passes the discovery assertion while concealing a tool. Conformance here is a claim an implementation makes about itself, made checkable — not an adversarial audit.
Interpretive independence. The independent implementation is independent of the reference code, not of its author's reading of the specification, because one person wrote both. Establishing it requires somebody else. It is the next validation threshold and the contribution this project most needs.
Agentic Datasets is an independent technical project initiated and maintained by Alexander Chernov (ORCID 0009-0007-3198-2712). The reference implementation is a release candidate: nothing is archived or minted yet, so a finding can still change the artifact rather than becoming errata against a published record.
The licences are chosen so that an implementation needs no permission.
© 2026 Alexander Chernov. This profile text is licensed CC BY 4.0. The mark is not: see the brand notes.