jev-blindspot
Prompt reviewer for Claude Code and Codex where Jev decides whether each prompt needs review before a light model lists blind spots.
Overview
jev-blindspot is a prompt reviewer that runs alongside Claude Code or Codex CLI sessions and shows reviews in a local web UI, or in the CLI via a Claude Code mod. Each time you submit a prompt, Jev decides in one call whether a review is worth doing; only then does a lightweight model, the agent's own headless mode, read the project and list what the request may have overlooked, such as missing context, unstated assumptions or the underlying problem.
It never blocks or edits the prompt and adds nothing to the agent's context; the prompt goes through as written and you decide whether to act. It installs from npm, stores the TypeSafe key in a permission-restricted config file, and hooks into every agent it finds; Codex requires trusting the hook once. The /blindspot command returns the UI URL without spending a model turn. The README documents what leaves your machine, costs, configuration and known limits, and it is MIT licensed.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Is this prompt worth a second look for blind spots? | yes · no |
How it works
- User submits a prompt to Claude Code or Codex
- Jev decides whether a review is needed
- If so, the agent's headless mode reads the project and lists blind spots
- Review appears in the web UI or CLI; prompt proceeds unchanged
Strengths
- Cheap gate avoids running reviews on every prompt
- Never blocks or alters the session
- Works with both Claude Code and Codex
Caveats
- Reviews use an additional model call when triggered
- Codex hook must be trusted manually
Getting started
`npm install -g jev-blindspot`, put TYPESAFE_API_KEY in ~/.config/jev-blindspot/env, then `jev-blindspot install-hook` and type /blindspot.At a glance
- Type
- CLI
- Status
- beta
- Language
- TypeScript
- Runtime
- Node CLI, agent hooks, local web UI
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
A side panel for Claude Code and Codex CLI that shows the blind spots of each prompt you submit: what the request would have needed to consider and shows no sign of. Jev decides, in one call per prompt, whether the prompt is worth a second look; only then does the agent's own headless mode read the project and return the items. Nothing blocks the prompt, nothing is added to the session.
logicrw/awesome-jev-projectsA prompt reviewer for Claude Code and Codex CLI. Finds the blind spots in each prompt you send and shows them in a local web UI, or in the CLI with the Claude Code mod.
hellogumbo/awesome-jevSide panel for Claude Code and Codex that asks what a prompt leaves unsaid before the agent acts on it.
MrJev/awesome-jev
Frequently asked questions
What does Jev decide?
Whether a submitted prompt merits a blind-spot review.
Does it change my prompt?
No. It never blocks, edits or adds to the session.
Where do I see reviews?
In a local web UI or in the CLI via a Claude Code mod.
Related projects
macOS menu bar screen-time app that asks Kev or Jev whether each screen breaks your rules and steps in with a pop-up.
ServicebetaPythonmacOS workspace for Codex with CodeGraph, Git and terminals, where opt-in Jev history recall finds past sessions and their decisions.
AppbetaNeovim popup showing jev-lens verdicts after Claude Code edits: whether to look, which files, and debris to strip.
ProjectbetaLuaPi extension that audits whether the rpiv-todo board matches visible work and nudges the agent when it drifts off-task.
PluginbetaTypeScriptExplainable day-trip planner where an LLM drafts the itinerary and Jev screens places, compares transport and reviews each step.
AppprototypeTypeScriptLocal natural-language task workspace where Jev, OpenAI or Gemini turn requests into view and task changes, with speed and cost comparisons.
AppprototypeTypeScript