Jev Review MCP
Local MCP server giving Claude Code, Codex, Cursor and OpenCode per-dimension Jev quality scores on their code while they work.
Overview
Jev Review is a local-first MCP server that gives coding agents structured software-quality scores as they work. The agent sends a focused diff and context to the single jev_review tool, and Jev returns typed Score, Choice and Noul decisions across independent dimensions such as correctness, complexity, changeability, modularity, tests and security. The server converts these into metric scores, confidence levels, coarse rubric hints and comparisons with the previous evaluation.
The intended loop is frequent checkpoints: a baseline after an implementation slice, then the agent forms a hypothesis about weak dimensions, improves the code, validates and rescores. Jev does not explain why a score is low; the coding agent must diagnose and change the code. There is deliberately no blended overall score, since per-dimension changes are more informative. The API key stays local, with no hosted backend, database or telemetry. It installs from GitHub via a plugins command for Claude Code, Codex, Cursor and OpenCode.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How does this change score on each quality dimension (correctness, complexity, modularity, tests, security...)? | — |
How it works
- Agent implements a coherent slice
- Agent calls jev_review with the diff and context
- Jev returns per-dimension scores
- Agent diagnoses weak dimensions and improves code
- Rescore and compare with the baseline
Strengths
- Fast scalar quality signals during development
- No overall score that hides dimension changes
- Local-only with no telemetry
Caveats
- Jev gives no explanation of low scores
- Diff and context are sent to the Jev API
- Not published on npm
Getting started
export JEV_API_KEY=..., then npx plugins add NiazMorshed2007/jev-review and choose your client.At a glance
- Type
- Service
- Status
- beta
- Language
- TypeScript
- Runtime
- Node MCP server (stdio)
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Local-first MCP: Claude Code, Codex, Cursor, and OpenCode get structured quality review from Jev while they write. Not the same project as [Jev Review](#applications) above.
AnotiaWang/awesome-decision-modelsA local MCP code-quality reviewer returning structured scores to coding Agents.
logicrw/awesome-jev-projectsLocal-first continuous software-quality review for coding agents.
cobanov/awesome-jevLocal-first MCP plugin for continuous software-quality review by AI coding agents, powered by Jev.
hellogumbo/awesome-jevLocal-first MCP server for continuous software-quality review by coding agents.
MrJev/awesome-jev
Frequently asked questions
What does Jev decide?
Scores for each software-quality dimension of the agent's change.
Does it fix code?
No. The coding agent diagnoses and changes code; Jev only scores.
Is there an overall score?
No, deliberately; dimension-level changes are reported instead.
Related projects
CLI and agent commands auditing a git diff against YAML standards packs, scoring each hunk with Jev so agents fix only flags.
Agent skillbetaTypeScriptCoding-agent hook that asks Jev one yes/no per project rule on every edit or turn and makes the agent fix any violation.
HookbetaTypeScriptCLI and MCP tool that finds hard-coded judgment calls in your code, scores how well Jev fits, and replaces strong fits with a fallback.
ProjectbetaTypeScriptMCP server giving coding agents fast typed decisions (ship gates, risk checks, file triage, page checks) as accept, review or escalate.
MCP serverbetaTypeScriptFuzzy linter hook for Claude Code and Codex where Jev checks each written file against team rules and reports violations at edit time.
HookbetaTypeScriptDrop-in proxy for Playwright MCP that adds Jev page triage, prompt-injection masking, snapshot pruning and risky-action gating.
MCP serverbetaTypeScript