JJEV·DIRECTORY GitHub agent pack connect your agent

** Jev CLI is out** <a:elmo_fire:1549940287235686410> `jev` puts TypeSafe's Jev model on the command line.

Homeby Nasr2026-09-18
** Jev CLI is out** <a:elmo_fire:1549940287235686410> `jev` puts TypeSafe's Jev model on the command line. **What it does** • `verify`: check claims against evidence, verdict + probabilities per claim • `screen`: flag prompt injection, filler, and off-topic text before an agent reads it • `classify`: single, multi, or hierarchical labels with confidence • `extract`: pull emails, amounts, dates, custom regex fields without hallucination • `match` : same / different / unclear for record pairs, with dedupe • `route` : pick a handler and fill its arguments in one call • `find` / `rerank` : rank up candidates against a plain-language query • `ask` : raw yes/no, choice, or score questions about any state • `compact` : shrink agent transcripts by dropping stale tool calls, kept lines stay verbatim • `batch` : run any of the above over thousands of rows with a concurrency pool Works with a TypeSafe key, OpenRouter, or Cloudflare Workers
Links
open source discussion

More in Home

I tried to create a latent context layer that is a warmed up version of a small LDA Topic model that is injectable into the context window via memories or system prompts.I made some CVSS classification scripts that use Jev to generate a CVSS vector for a given software vulnerability description.Jev can be used for profiling players, which is great for organizing playtests and checking balance changes across different player groups.