AI Model Routing Rule Checker
Compare recorded provider choices with declared allowlists, dated capability snapshots, retention requirements and region rules without routing requests.
Loading tool…
How to use AI Model Routing Rule Checker
A routing setting is useful only if the observed provider choice agrees with the intended constraints. Audit supplied request records against dated provider evidence and preserve missing facts as unknown.
- Copy neutral request/provider identifiers and the exact dated provider evidence snapshot applicable to each record.
- Declare your allowed providers, required capabilities, optional region and evidence freshness policy.
- Investigate contradictions and unknowns with actual contracts, logs and live authorized tests before trusting a deployment.
Good to know
- Provider facts are entered declarations, not live capabilities. Stale, missing or future evidence does not establish a requirement; a known disallowed provider or declared no remains a contradiction.
- An exact region label is not verified data residency. No route, fallback, retention contract, endpoint, API or traffic is contacted or changed. Maximum evidence age is 3,650 days.
- Each input field is limited to 8,000 UTF-16 characters. CSV needs the exact displayed header and at most 100 data rows. Quoted commas, doubled quotes and quoted newlines are supported. Duplicate identities are refused unless the tool explicitly reports them as findings.
- Markdown and CSV are editable review artifacts, not certificates. Spreadsheet-like text is apostrophe-prefixed in CSV to reduce formula execution risk; check the actual importer before sharing. Decimal ratios are rounded to at most 12 places.
- Runs only on the records you enter, locally in your browser. No AI call, provider connection, account inspection, upload, automatic outreach or saved input. Use neutral identifiers rather than personal or secret content.
Research behind this workbench
These cases explain the business mechanism. The workbench is ToolJovia's independent adaptation, not a claim about the company's internal workflow or an endorsement.
Common questions
Does a supplied-fact match certify zero retention?
No. The result merely finds no contradiction in the supplied declarations. Applicable provider contracts and actual configuration still need review.
Why use a separate evidence ID per recorded request?
Provider facts can change or be incomplete. The evidence ID makes the specific snapshot explicit instead of silently applying today’s fact to a different observation.
Keep going with related tools
All developer →Model Prediction Cost Ledger
Reconcile observed model prediction units and explicit billing states while keeping unknown charges separate from a known-cost subtotal.
DeveloperPrompt Release Version Checker
Match trace version references to approved prompt-release intervals and flag overlaps, missing approvals and fingerprint mismatches.
CalculatorsApp Runtime Budget Guardrails
Separate runtime billing dimensions, included allowances and fixed costs, then calculate budget headroom and conditional unit limits.
DataDashboard Viewer Data Check
Compare explicit viewer row entitlements with observed dashboard fixture rows, keeping unknown scopes distinct from empty approved access.