Jev governance guide
Governance for Jev is change control on questions, model pins, and who can enable auto-act. The model will not govern itself.
This page is an independent field-guide note, not a GRC product and not a TypeSafe playbook. jev.pro is not affiliated with TypeSafe AI. 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.
Minimum viable governance
- Named owner per question pack
- PR review on criteria (TypeSafe skill advice: humans edit)
- Production pin ≠
jev-latest - Dual-control to lower floors
- Quarterly sample of auto-actions
Criteria are descriptions a person could apply. Changing them is a new decision contract even if the pin stays. Replay gold before you promote. Atomic questions live in code; one POST /v1/systemone can carry many. First-party API is that POST plus GET /v1/models.
Choice maps need other / none when coverage is incomplete (max 255). Score has 2–10 ordered levels; keep arithmetic in code; do not treat score as dollars. Noul is P(yes); 0.5 is uncertain, not medium. Confidence is peakedness of Choice/Score; Noul has none. Dual-control to lower floors exists because a copied τ is how irreversible auto-act sneaks in.
If you are migrating a prompt-owned policy into questions, use the migration guide: shadow mode, then review-heavy, then narrow auto. Governance is who signs each of those cuts.
Evidence
Traces + question git history are the workpaper. See evidence templates and the decision trace glossary. The action space (including abstain) is what you are governing, not the model’s personality.
RACI sketch (unofficial)
| Activity | Platform | Product | Risk/Legal | Jev (the model) |
|---|---|---|---|---|
| Edit criteria | R | C | C | — |
| Lower auto floors | A | C | C | — |
| Irreversible side effect | I | A | C | never |
| Vendor DPA | I | I | A | — |
R/A/C/I are yours to assign. The last column is the point. Public docs emphasize request-shaped criteria, not a BRMS UI. Your control owner signs. Never Jev.
Security and legal interrogations: security guide, questions for legal, vs compliance checklists. Implementation and launch still apply: implementation checklist, launch checklist. Fail-open auto-act is an anti-pattern. Scope: when to use Jev.
What this page does not claim
- Not ISO/SOC certification.
- Not official TypeSafe governance documentation.
- Completing the RACI table is not an attestation.
FAQ
Does TypeSafe offer a policy console?
Public docs emphasize request-shaped criteria, not a BRMS UI.
Who signs?
Your control owner. Never Jev.
Can we lower floors in the same PR as a criteria tweak?
Prefer not. Dual-control exists so a wording change does not silently widen auto-act.
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 compliance checklists. Canonical: https://docs.typesafe.ai.