Glossary: Audit trail
Audit trail is the durable, reviewer-facing record of who asked what, which versioned model answered, the full distributions, and which verdict your code took. It is the production name for stored decision traces plus identity and policy metadata. Jev (TypeSafe System One) will not write “because…” into the trail. Do not claim a later paragraph is Jev’s reason.
This glossary entry is for builders who already heard of Jev / System One and need a citable definition before they assume TypeSafe stores their log forever. jev.pro is an independent unofficial field guide — not affiliated with TypeSafe AI. Confirm API fields on docs.typesafe.ai. We do not sell or issue API keys.
How it shows up
Persist response.model, usage, answers, filtered state or a hash, and the actor. If compliance wants a paragraph, generate it later and label it assistance. Hash if the payload is sensitive; keep enough fields for a replay. Replay is your POST, not a hidden endpoint.
Do not assume TypeSafe stores your audit trail. Retention and ZDR are legal topics — read TypeSafe legal pages and your MSA. Keep a copy you control. Log the resolved id even if you sent jev-latest (currently jev-1.13.0 on the vendor snapshot we read). Store Choice/Score probabilities and confidence; store Noul as noul only. Calibration reviews need those rows. Schema-safe ≠ factually correct.
The trail should include the threshold version and the action space member committed. RLCD is not a column TypeSafe emits.
Contrast
Audit trail vs decision trace. Trace = one evaluation’s record. Audit trail = durable store + who + policy version + retention.
Audit trail vs generated essay. Jev does not produce reasoning text. A later LLM paragraph is assistance.
Audit trail vs TypeSafe logs. Official retention is a legal/product question. This site does not invent it. Keep your own copy.
Audit trail vs signal. Signals are answers. The trail is answers plus provenance.
Common mistakes
- Assuming TypeSafe is your system of record.
- Storing only the verdict.
- Attributing a generated “because” to Jev.
- Logging the alias but not
response.model. - Treating a schema-safe row as proven fact in an audit packet.
See also
FAQ
Does TypeSafe store my audit trail?
Do not assume that. Retention and ZDR are legal topics — read TypeSafe legal pages and your MSA. Keep a copy you control.
Hash or raw state?
Hash if the payload is sensitive; keep enough fields for a replay. Replay is your POST, not a hidden endpoint.
Can Jev write the audit narrative?
No. Persist typed fields. Label any later prose as assistance.
What identity belongs on the row?
The actor who triggered the evaluation, the service version, and the policy/threshold id — not a fabricated Jev “reasoner id.”
Limits
This is a jev.pro working definition unless TypeSafe uses the same word as an API field. We do not invent search volume or retention SLAs. Schema-safe answers are not automatically factually correct.
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.