scriptkittyos & labs

research

Measured, not declared.

The benchmark, its frozen methodology and its per-trial results are public, so an assessor can examine the evidence before the system is installed. Papers are deposited with DOIs.

Papers.

TitleTypeIdentifier
The Model Proposes, the System Authorizes: An Authority Control Plane for AI Agents on the BEAMposition paperconcept DOI 10.5281/zenodo.21754762
Authority-Bound Agentic Execution: Measuring Unauthorized Effect Under Adversarial Loadpaperconcept DOI 10.5281/zenodo.21755869
Schrödinger's Cyber Security Framework: Vulnerability as an Observer-Dependent Quantitydepositconcept DOI 10.5281/zenodo.22116617
Security Challenges in AI Agent Deployment: Insights from a Large Scale Public Competition (co-authored; the ART agent red-teaming benchmark)paperarXiv:2507.20526

Benchmarks.

HolyTrinity-Benchmark v1.2 (3 August 2026, CC BY 4.0). An adversarial security benchmark for agent authorization. 61 attack trials (73 with 12 controls); 0 unauthorized effects; per-trial 95% bound ≤5.9%, cluster-aware ≤32.4%; the violating action driven in 57 of the 61 attack trials; median governed-action latency about 19 ms. Two standalone verifiers, Elixir and dependency-free Python, with defined exit codes.

Requisition is known internally as HolyTrinity. The system under test is Requisition, measured at a private commit under its working name; the repository calls it "the Trinity control plane". It is not the open-source Trinity agent.

v1 is frozen. The numbers describe a superseded build.

Requisition Bench: 28 pre-registered scenarios comparing Requisition with the Dogwood policy engine. Every result matched its pre-registered prediction: 14 agree, 8 where Dogwood allows and Requisition refuses, 6 where Dogwood denies and Requisition allows. Requisition re-checks standing and holds at execution; Dogwood replays a trace, so some of those verdicts are the point of its design. Counts only, a small set, under a policy pack Script Kitty wrote rather than one authored natively for Dogwood. Since v0.2.0 each row carries a signed receipt chain verifiable offline with the HolyTrinity-Benchmark verifier. The findings · repository.

ONE-Bench, the Open Neuro-Execution Benchmark for brain-computer-interface control pipelines under normal, degraded, adversarial and out-of-distribution conditions. Research, in preparation; not an offered deployment.

Independent review.

Requisition Bench's expected verdicts were reviewed by Zaid Hasan Khan, an independent healthcare AI reliability evaluator. His work tests whether an AI workflow keeps the facts it needs when context changes: missing information, stale evidence, contradictions, handoffs and escalation.

28 of 28 agreed On 26 September 2026 he agreed all 28 expected verdicts, 0 disputed. The review file is bound to exact bytes, so an edit to any reviewed file invalidates it by name.
His practice Controlled synthetic cases, with expected behavior locked before outputs are observed and key cases repeated to check that findings recur. Behavioral reliability evaluation, not clinical validation or certification. Based in India; works remotely.

How the evidence is kept honest.

Independent examiners hold a frozen specification of the authority boundary and test against it. The goalposts do not move.

No change lands without a failing test first. A criterion that cannot fail is not a criterion, and a check that cannot fail is not a check.

Cite.

Each repository carries a CITATION.cff. Cite the benchmark by URL and tag; it has no DOI of its own. Cite papers by concept DOI unless a specific version is the point.