Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

jev-kit

Codex skill and MCP for advisory Jev decisions in Java/Maven backend work: file relevance, change risk, SQL risk and more.

MCP serverbeta
View on GitHub ↗

Overview

Jev Kit (chrisstarr7) is a Codex skill plus MCP integration for bounded, advisory decisions in Java backend development. Jev classifies a small set of engineering questions while the coding assistant still owns reading code, implementation, review and delivery. Decisions include narrowing a large candidate file list, choosing review depth from change risk, and flagging SQL and migration risk, among others in a decision pack.

The assistant reads relevant code, decides whether one bounded classification would help, sends short redacted context to the jev_ask MCP tool at most once per user turn, and accepts the answer only with confidence of at least 0.75 and agreement with code evidence; otherwise it continues normally. Jev cannot write code, approve production changes, waive tests or bypass reviews. The npx installer adds the skill, decision pack and MCP entry for Maven projects and never writes the key to disk. Documentation is in English and Chinese.

What the decision model decides

TypeQuestionAnswers
ChoiceWhich of these files are relevant to the task?—
ScoreHow risky is this change, and what review depth fits?—
ScoreHow risky is this SQL or migration?—

How it works

  1. Install in a Maven project with npx
  2. Codex loads the skill when the task matches
  3. Optionally send redacted context to jev_ask once per turn
  4. Use the answer only if confidence is at least 0.75 and agrees with code

Strengths

  • Strictly advisory with clear limits
  • Redacted context
  • Java/Maven-specific decision pack

Caveats

  • Installer runs the current default branch; review before running
  • Live requests need uvx and a key

Getting started

From a Maven project root: `npx -y chrisstarr7/jev-kit`, then restart Codex.

At a glance

Type
MCP server
Status
beta
Language
TypeScript
Runtime
Codex skill and MCP
Has code
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

What does Jev decide?

File relevance, change risk, SQL risk and similar bounded engineering questions.

Can it approve changes?

No, it is advisory only.

What confidence is required?

At least 0.75 plus agreement with code evidence.

Related projects

  • agent-fastpath

    MCP server giving coding agents fast typed decisions (ship gates, risk checks, file triage, page checks) as accept, review or escalate.

    MCP serverbetaTypeScript
  • Hunch (Kelbie)

    Semantic code search and plain-English review rules that send repository chunks to Jev and return scored source locations.

    ProjectbetaTypeScript
  • Antigravity-mcp-semantic-search-with-TypeSafeAi

    MCP server that finds relevant code with Jev Choice and Noul checks instead of grep sweeps, and audits diffs against task intent.

    MCP serverprototypePython
  • jev-map

    Maps each Python function to the tests that run it; Jev scores candidate function-test links that static analysis misses.

    ProjectbetaPython
  • jev-debug

    Locates the file a bug report is about, or flags defects in a diff, by asking Jev yes/no over small source batches.

    ProjectresearchPython
  • agent-dispatcher

    Claude Code and Codex skill that routes a task to one of 27 specialist roles, with an optional Jev engine for catalog selection.

    ProjectbetaPython