Know what the simulation records.

How scripted aids participate

Northstar uses fixed, deterministic, package-frozen research, analysis, and review responses. They are not live general-purpose AI and cannot establish prompting skill, AI literacy, or human-AI complementarity. Accepted request metadata, exact fixed output, source boundary, configuration, request time, and release time remain traceable. Scripted aids hold no approval or rating authority.

What reviewers may conclude

Trained reviewers may make bounded, rubric-linked statements about this work episode. Ratings retain their evidence references, confidence, comments, version, and disagreement. TENX ARENA does not produce a Hire or Reject result, candidate ranking, fit score, synergy score, or prediction of future performance.

Accessibility and accommodations

Your invitation is bound to an accommodation form. Server timing and opportunity windows use that form. Keyboard operation, visible focus, responsive text, reduced motion, and assistive-technology compatibility are treated as measurement conditions. The system never interprets missing opportunity as poor performance.

Storage and retention

A session starts only when the configured transactional store is available. Role-scoped access, expiry, revocation, audit records, correction requests, and deletion-request status are part of the governance foundation. This research version has no approved operational retention schedule, so it must not collect real participant data. Bearer credentials are excluded from evidence exports.

Research is separate / TENX-RESEARCH-1.3

Optional research use is limited to studying this instrument's opportunity delivery, event capture, replay integrity, rating process, reliability, and validity. It is not permission to make a hiring decision or train a general employment model.

The metadata-only research export replaces the operational session identifier and session-scoped record identifiers with domain-separated, package-bound server-keyed pseudonyms. It may include fixed package and source-capture metadata; event types, relative elapsed times, and hashes; non-protected object and opportunity states; artifact hashes, lengths, and lineage; consent status; and categorical ratings, evidence references, disagreement, and provenance. It excludes artifact text, source-snapshot content, event payload bodies, rating comments, adjudication rationales, exact accommodation identifiers, administration duration multipliers, adjusted opportunity windows, administration end times, the export time, and absolute participant, event, artifact, consent, rating, adjudication, and verification times. Withheld fields are marked as not included rather than represented as empty.

Pseudonymization is not anonymity. Sequence, relative timing, the occurrence of an accommodation-related event, fixed source-capture metadata, rating metadata, and retained exact content or provenance hashes may still create re-identification or linkage risk when combined with other records, even though the protected details listed above are excluded. The keyed pseudonyms reduce direct joining by operational identifiers but do not remove this residual risk.

Research access is permitted only to people named by an approved deployment-specific policy. That policy must state the responsible organization, research purpose, authorized recipients, safeguards, retention limit and deletion process, participant contact, and any additional risks before this choice may be enabled.

Declining or withdrawing optional research use does not affect simulation completion. While this authenticated session control remains available, you can withdraw in the completion screen and cannot re-enable consent under this notice version. After access expires, withdrawal requires the deployment-specific contact and workflow. If that policy, contact, retention limit, or workflow is not configured, research consent and production participant issuance must remain disabled.

Content hash: sha256:4827af528962edf6b9c61da30b6409111bd011ec81c8be1ce4a53010488f2ba4

The research download is structurally metadata-only. It uses server-keyed package-bound pseudonyms for the operational session and session-scoped records, retains relative elapsed time rather than participant-linked wall-clock time, and omits the export timestamp. Artifact and source content, event payload bodies, free-form reviewer text, evidence-reference notes, exact accommodation identifiers, duration multipliers, adjusted opportunity windows, and administration end times are omitted. Exact content and provenance hashes remain a stated linkage risk. Explicit disclosure markers distinguish withheld data from empty data.

Questions, correction, and withdrawal

The submitted-session screen provides a one-way, server-recorded research-withdrawal control while the authenticated session remains available. After that access expires, a deployment-specific contact and external workflow are required. Neither a contact, approved retention limit, nor that external workflow is configured in this engineering version, so production participant issuance and research consent must remain disabled. Corrections append provenance records rather than silently rewriting prior accepted events or ratings.