Gaze-in-the-Wild candidate ProcessData preflight

This governance layer screens one explicitly selected file inside a quarantined Gaze-in-the-Wild recovery candidate against GazeForge's ProcessData structural contract.

It exists to answer one narrow question: does this exact reviewed file have the structure required by the ProcessData-side adapter? A positive answer is not evidence that the candidate is authoritative, byte-equivalent to the historical external distribution, licensed for analysis or redistribution, complete, or empirically eligible.

Required sequence

1. Build and validate the generic recovery-candidate review with build_gaze_in_wild_recovery_candidate_review().
2. Keep every generic inventory file role unclassified.
3. Explicitly select one reviewed inventory path for structural screening.
4. Call build_gaze_in_wild_candidate_processdata_screen().
5. The screen re-verifies the complete candidate tree, binds the selected path to its reviewed SHA-256 and byte size, and runs preflight_gaze_in_wild_processdata() on that exact file.
6. Persist a screen only with write_gaze_in_wild_candidate_processdata_screen(), which re-runs the binding checks and requires the output to remain outside the candidate tree.

What a passing screen establishes

A passing record establishes only that:

• the recovery review was valid and still matched the complete candidate tree at screening time;
• the operator-selected path identified exactly one reviewed inventory file;
• that file's current bytes matched the reviewed SHA-256 and size;
• the file passed the ProcessData structural checks required by the current adapter; and
• the screen is reproducibly bound to the exact recovery-review and tree fingerprints.

The generic recovery inventory remains unchanged. In particular, the selected file's generic role stays unclassified; a passing screen does not rename or promote it to an authoritative ProcessData distribution member.

Scientific and rights boundary

The screen keeps the candidate in quarantined status and cannot establish any of the following:

• source authority;
• original-distribution or exact-copy identity;
• dataset-file rights, analysis-use permission, or redistribution permission;
• participant, trial, task, or labeller mapping;
• coordinate-unit semantics;
• corpus-wide sampling cadence or published acquisition cadence;
• recovery of a separate LabelData object or independent labeller streams;
• quarantine-exit authorization;
• source-audit readiness or empirical evidence eligibility;
• human-human agreement, participant-disjoint model validation, cross-dataset performance, GP3 validity, or Frozen Evidence performance claims.

ETG.Labels observed inside a ProcessData object is therefore never interpreted as recovery of the independent labeller LabelData files.

Relationship to quarantine exit

This screen is intentionally orthogonal to the quarantine-exit authorization contract. Structural compatibility can help reject an incompatible candidate early, but it does not satisfy the independent authority, exact-copy, rights, reuse-terms, or analysis-permission evidence required for an authorized quarantine exit.

The unresolved Gaze-in-the-Wild milestone therefore remains unchanged: obtain an authoritative corpus/source copy and explicit dataset-file reuse terms before any empirical source-audit gates can open.