claude-jev
Claude Code plugin where Jev decides and Claude writes: sort piles of text, calibrate thresholds, route prompts and gate Bash commands.
Overview
This jev plugin by zachshallbetter puts Jev inside Claude Code with an explicit division of labour: Jev decides, Claude writes. /jev:jev sort, ask and route sort, triage, flag or score piles of text such as emails, tickets, invoices, logs or folders, using built-in presets or your own questions, with a dry run first, consent before anything leaves the machine, and an "unsure" pile you judge yourself.
/jev:jev calibrate measures whether Jev's confidence can be trusted on your own labelled examples before you rely on a threshold, shipping a 66-message household set and an HTML report. Two opt-in features are off by default: a prompt router where Jev sizes each prompt and suggests one of four bundled helper agents by model tier, and a permission gate that auto-approves Bash commands Jev is sure about. A reference guide covers building Jev into code for routing, guardrails, reranking and tool-call gating.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which pile or category does this item belong to? (sort/triage) | — |
| Choice | Which helper agent tier should handle this prompt? | jev-haiku · jev-sonnet · jev-opus · jev-fable |
| Yes / no | Is this Bash command safe to auto-approve? | yes · no |
How it works
- Choose a sort, ask or route preset or write questions
- Dry run and consent before sending
- Jev judges each item; uncertain ones go to an unsure pile
- Optionally calibrate thresholds on labelled examples
Strengths
- Consent and dry run before data leaves
- Built-in calibration with a report
- Risky features off by default
Caveats
- Auto-approving commands carries risk if enabled
- Needs a TypeSafe key
Getting started
/plugin marketplace add zachshallbetter/claude-jev then /plugin install jev@claude-jevAt a glance
- Type
- Plugin
- Status
- beta
- Runtime
- Claude Code plugin
- Has code
- yes
Use cases and integrations
How curated lists describe it
Jev (TypeSafe System One) as a Claude Code plugin: typed decisions for sorting…
wh000wh000/awesome-jev-live
Frequently asked questions
Can I check Jev's calibration first?
Yes, /jev:jev calibrate measures it on your own labelled examples.
Is the permission gate on by default?
No, both the router and permission gate are opt-in.
What happens to uncertain items?
They go to an unsure pile for you to judge.
Related projects
Claude Code plugin and CLI for the Jev funnel: judge items in bulk, settle confident ones in code, send only uncertain ones onward.
PluginbetaPythonMultiplayer GitHub triage dashboard where Jev answers kind, severity, urgency, duplicate and next step per issue.
AppbetaCLI and Claude Code skill that turns one typed decision into a fine-tuned, calibrated, evaluated Laya checkpoint you can serve.
ModelbetaPython- SkillRanker★ 134
Rust CLI that ranks which agent skills fit the next step from live session context with Jev, can abstain, and hooks into Claude Code.
CLIbetaRust Janus (formerly calibre): measures on your data the confidence threshold at which a small model like Jev should defer to a larger one.
ProjectbetaPythonRecursive language model layer for coding agents that keeps inputs in a local evaluator, with optional Jev judge_many for typed calls.
ProjectbetaRust