Skip to main content
TENXARENA

Admin workspace / release declarations

Open only what the evidence and controls support.

This page shows declarations shipped with this release and safe observations from its render environment. It is not live monitoring or an operational readiness decision.

Controlled synthetic participant and reviewer adapters are available only when their explicit release gates are enabled and their scoped credentials are valid. Real candidates and hiring use remain blocked.

Release declarations. These values are not live telemetry.

21canonical migrations in the accepted operational closure
0qualified live runtime models in this release
$0live-runtime call budget for this readiness check
Gatedsynthetic participant and reviewer adapters only

Canonical foundation declarations

Database baseline

Repository release record

Historical releases remain preserved. The current source contract is v2.2.15; twenty-one canonical migrations cover runtime, governance, controlled synthetic persistence, lifecycle operations, application-owned authentication, and authenticated database requests. A source contract is not a publication approval or a statement of live database health.

Release declaration
Package and event contracts

Source contract scope

The current frozen source manifest binds the exact v2.2.15 interpreter, content addressing, package binding, event chains, artifact revisions, provenance, rating, adjudication, and employer report projection. Historical evidence contracts are preserved, not rewritten or silently upgraded.

Release declaration
Database authority isolation

Four restricted authorities

Control plane, participant runtime, reviewer evidence, and governance use separate restricted database roles with role-scoped health assertions.

Synthetic architecture
Deployment health

Not reported here

This page does not request or display live deployment health. A public health response is not a protected operational readiness decision.

Use authorized monitoring

Governance adapter declarations

Named identity provider

This real-use readiness marker remains closed. The controlled synthetic QA adapter path does not satisfy this production gate.

not integrated
Tenant and study governance store

This real-use readiness marker remains closed. The controlled synthetic QA adapter path does not satisfy this production gate.

not integrated
Resource-bound runtime authorization

This real-use readiness marker remains closed. The controlled synthetic QA adapter path does not satisfy this production gate.

not integrated
Shared atomic abuse control

This real-use readiness marker remains closed. The controlled synthetic QA adapter path does not satisfy this production gate.

not integrated
Purpose-separated keyring

This real-use readiness marker remains closed. The controlled synthetic QA adapter path does not satisfy this production gate.

not integrated
Independent integrity anchor

This real-use readiness marker remains closed. The controlled synthetic QA adapter path does not satisfy this production gate.

not integrated

OpenRouter policy lab

Inspect the new capability laboratory, workload gates and current dated screening. The older three-purpose policies below remain inactive historical declarations.

The adapter is server only and designed to fail closed. It needs a valid server key, positive call, session, and study budgets, plus an unexpired approval for the exact evaluation. Every listed model is unqualified for TENX use.
Authoring draftunqualified

anthropic/claude-sonnet-5-20260630

Exact endpoint: amazon-bedrock/global. Automatic provider and model fallback are disabled.

  • Zero data retention required
  • Provider data collection denied
  • Strict JSON Schema requested and checked again on the server
  • One bounded repair attempt on the same route
  • Independent endpoint attestation required, with no production adapter in this release

Recorded price snapshot from 2026-09-06: $2.00 per million input tokens and $10.00 per million output tokens. Example estimate for 8,000 input and 2,000 output tokens: $0.0360 before possible platform fees and cache effects.

Actor simulationunqualified

openai/gpt-5.6-luna-20260709

Exact endpoint: azure/eu. Automatic provider and model fallback are disabled.

  • Zero data retention required
  • Provider data collection denied
  • Strict JSON Schema requested and checked again on the server
  • One bounded repair attempt on the same route
  • Independent endpoint attestation required, with no production adapter in this release

Recorded price snapshot from 2026-09-06: $0.22 per million input tokens and $1.32 per million output tokens. Example estimate for 3,000 input and 500 output tokens: $0.0013 before possible platform fees and cache effects.

Evidence summarizationunqualified

google/gemini-3.1-flash-lite-20260507

Exact endpoint: google-vertex/global. Automatic provider and model fallback are disabled.

  • Zero data retention required
  • Provider data collection denied
  • Strict JSON Schema requested and checked again on the server
  • One bounded repair attempt on the same route
  • Independent endpoint attestation required, with no production adapter in this release

Recorded price snapshot from 2026-09-06: $0.25 per million input tokens and $1.50 per million output tokens. Example estimate for 20,000 input and 1,000 output tokens: $0.0065 before possible platform fees and cache effects.

Server key observation

Render environment only

This safe boolean shows whether the render environment exposed a server variable. It does not check key validity, account settings, endpoint access, or readiness.

Observed, use blocked
Budget input

Page calculation only

This page passes zero budgets to its local check. It does not inspect a shared spend ledger.

Page input: zero
Model evaluation record

Release declaration

The release work records no model inference. Public benchmarks and dated provider metadata are screening evidence only.

Release record: not performed

Evidence distinctions

External research support

May motivate a product mechanism. It does not validate TENX.

Occupational evidence

May inform a planning draft. It is not an employer fact or score weight.

Benchmark evidence

May screen a technical option. It is not candidate-assessment evidence.

Internally tested engineering behavior

Shows software behavior under tested conditions only.

Design hypothesis

A proposed mechanism awaiting user and research evidence.

Governance requirement

A hard control required before an intended use can open.

Empirical TENX evidence

Requires a prospective TENX study with a stated design and uncertainty.

Not yet validated

The required inference evidence does not exist.

Controls still required for real use

  • Named staff identity with strong authentication, stable participant identity, and organization membership.
  • Independent operational verification of tenant-scoped persistence and authorization on every package, invitation, session, rating, export, and governance request.
  • Preserve and independently exercise the four restricted database authorities, separate connection secrets, route-scoped grants, and authority-specific health checks.
  • A shared atomic rate limiter, versioned purpose-separated keyring, and independently credentialed integrity anchor.
  • Operational consent withdrawal, data subject fulfillment, retention, deletion, legal hold, incident suspension, and restoration exercises.
  • Independent subject matter review, manual accessibility evidence, rater reliability, fairness, criterion, and intended-use validation.