Ask Jev, TypeSafe's decision model, to pick one option, place something on an ordered scale, or answer yes/no, with calibrated probabilities. Several questions can go in one call. Read-only: nothing is saved unless you choose to. Bring your own TypeSafe API key (console.typesafe.ai).
What your agent can do with it
- jev_apiAuthenticated HTTP access to TypeSafe's API (https://api.typesafe.ai, POST /v1/systemone). The dispatcher injects the vaulted key as `Authorization: Bearer <key>`; credentialRef i…
- jev_chooseAsk Jev to pick ONE option about some content, with a probability per option and Jev's confidence. Read-only: saves nothing. Args: { state: string|object (the content), question: …
- jev_scoreAsk Jev to place content on an ordered scale (lowest level first). Read-only. Args: { state, question: string, levels: string[] (2+, ordered) }. Optional model: a Jev model name o…
- jev_noulAsk Jev a yes/no question about content; returns the probability of yes (no confidence field). Read-only. Args: { state, question: string }. Optional model: a Jev model name or al…
- jev_askAsk Jev several questions about the same content in ONE call (cheaper and faster than one call each). Read-only. Args: { state, questions: { name: { type: 'choice'|'score'|'noul',…
- jev-routing-policyWhen to route a decision through Jev, which verb to use, and what to do with a low-confidence answer. Pure know-how.
What you will need
Asked for at install. Stored in your pod, never on our servers.
- TypeSafe API keyYour TypeSafe API key from console.typesafe.ai. Stored encrypted in the pod vault; never in the template.
- {{name}} Jev API keySecretJev API key for the jev_api connector.