Questions for security reviewers
Security reviewers should assume hostile tickets and egress to a third party. This list is a start, not a pentest. Jev is a typed hop, not your SIEM, WAF, or malware lab.
This page is for security reviewers sitting with implementers. jev.pro is an independent unofficial field guide — not affiliated with TypeSafe AI. Open-cluster notes are not a TypeSafe playbook. Confirm facts on docs.typesafe.ai. We do not sell keys.
Open-cluster pages are independent field-guide notes. Replicas and third-party interfaces mentioned anywhere on jev.pro are not Jev and not endorsed.
Ask the implementers
Keep this interrogation as-is; expand around it, do not replace it:
- Where do keys live? Rotation?
- Which fields leave the VPC?
- Injection test set?
- Can a ticket rewrite instructions?
- Privileged auto-act on timeout?
- Log redaction?
- Door extra statuses handled (402, verification)?
- Who can lower floors?
If the team cannot answer (1) and (2), stop. jev.pro never issues keys. First-party evaluation is POST /v1/systemone plus GET /v1/models. Marketplace doors add their own error codes — treat 402 and verification as callouts, not retries of the same body.
Question (4) is a state-filter question. Backtick the fields you meant; do not POST the raw ticket if it can contain “ignore previous instructions.” Official jaggedness notes include adversarial content — confirm on docs.typesafe.ai. Do not put malware samples in state. Extract indicators in a sandbox first.
Residual risk you will still own
Hosted model, adversarial state, gateway extra error codes, logs that contain ticket text, humans who rubber-stamp high confidence. Jev does not remove those. It adds a typed hop.
Confidence is peakedness of a Choice/Score map, not a security control. A Noul has no confidence field; 0.5 means uncertain, not “medium risk.” Fail closed on timeout. Privileged auto-act when the vendor is unreachable is a product decision you will own in the post-incident review.
Jev is only a control as a component inside your control. See vs security automation and the longer security guide. Legal questions live on questions for legal.
How this review sits next to other gates
Wire-up habits: implementation checklist. Ship gates: launch checklist. Architectures that skip traces or fail-open: anti-patterns. Scope: when to use Jev — Jev is not a WAF. Persist a decision trace on production calls you might have to explain.
What this page does not claim
- Not an attestation.
- Not a replacement for your threat model.
- Not official TypeSafe security documentation.
FAQ
Is Jev a control?
Only as a component inside your control.
Malware samples in state?
Don’t. Extract indicators in a sandbox first.
Does a high confidence score mean the ticket is safe?
No. Confidence is distribution peakedness on Choice/Score. It is not a malware verdict.
Disclaimer
This is an independent unofficial site and is not affiliated with TypeSafe AI; official documentation is available at https://docs.typesafe.ai. Never treat jev.pro as TypeSafe official documentation. We do not sell, issue, or proxy API keys.
Open-cluster pages are independent field-guide notes. Replicas and third-party interfaces mentioned anywhere on jev.pro are not Jev and not endorsed. Hub: Open. Siblings: security guide, questions for legal, vs security automation. Canonical: https://docs.typesafe.ai.