Skip to content
Benchmarks

EVALUATION METHOD

Holdout

Frozen cases run in isolation against a frozen candidate. Only aggregate counts cross the publication boundary, so there is no case to open. Execution-qualified means the infrastructure ran its contract, not that a scanner detects well.

  • Qualification run 19510b39 · 2026-09-29
  • Source Frozen report in the repository

1 · How it runs

Did the frozen candidate run its holdout cases?

  1. InputFrozen public control cases, sealed at execution.
  2. ChangeNone.
  3. CheckEach scanner is read against the cases once. Counts are kept per stratum: the expected kind and tier.
Cases
12
Variants
12
Scanners
3

2 · Recorded now

What the holdout run recorded

Execution qualified (engine-conformance), run 19510b39 on 2026-09-29. It records that the infrastructure executed its contract. It makes no detection-quality or support claim.
Holdout checks per scanner and stratum
Stratumredact-secret0.1.0-beta.11gitleaks8.30.1trufflehog3.97.4
Must not flag, tier T30 of 60 of 60 of 6
Must redact, tier T10 of 60 of 60 of 6

Checks that did not hold, of those scored for that scanner.

3 · How to read it

Reading holdout

  • This page reads a qualification run, which is a separate snapshot from the discovery run the other methods read.
  • Public controls cannot establish independent detector performance, and the aggregate has no detector attribution or unique affected cases.

4 · Exact inputs

The corpus and the candidate

A holdout case cannot be opened. These are the aggregate facts the report publishes.
Corpus
engine-v1-public-controls, revision 1
Purpose
public-conformance
Lifecycle
sealed-at-execution
Independence
public-control
Methodology
frozen-candidate-canonical-cases-aggregate-only
Cases
12 cases, 12 variants, 0 generation errors
Plan
11390c307f4d
Full hashes of the corpus, the plan and the candidate
{
  "corpusHash": "821e8925e1534d06b7077ca5da7817b7241fe74784965f8ab63cf3ef40728b25",
  "seedHash": "f21cababc3d97ac61c51241c50b897fcad8f7514bdf3269110d5842ec3e227d3",
  "planHash": "11390c307f4d8bd507add26a821a2991dc2f5debb5f48795fb7329f4c7577aa9",
  "candidate": {
    "sourceHash": "94ac077df27ec4687593901405391302b17d2d40752b798ae818dbe9ccbbe949",
    "lockHash": "06a6ba659e9ae2d42ee49f0f11e13d682ed242539c89b6bb455c29cdb089c51f",
    "candidateArtifactHash": "9fba9d6b5a20c906c6c3abeaad247a70b7f2dfef15af965e7600716e65c15f8c"
  }
}