openjev-mcp
MCP server giving agents typed Jev decisions (ask, choice, score, noul, batch) with calibrated probabilities via the OpenJEV API.
Overview
openjev-mcp is an MCP server that lets AI agents get typed answers with calibrated probabilities instead of prose for questions like classify this, how severe is this, or is this urgent. It is powered by Jev through the public OpenJEV API: one context and any number of independent questions per call, with local request checks, retries for transient failures and validation of every answer before the agent sees it.
Tools include jev_ask, jev_choice, jev_score, jev_noul and jev_batch. The README lists uses such as support triage, content moderation, prompt-injection screening, agent routing, PR risk, lead qualification, LLM-as-judge evals, mapping free text to one of 200 SKUs and backfilling queue triage. Recorded transcripts show DeepSeek Harness triaging an electric co-op's help desk. One-click install links exist for Cursor and VS Code, and it runs with npx -y openjev-mcp and an OPENJEV_API_KEY.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which team owns this ticket? | — |
| Score | How harmful is this message? | none · mild · serious |
| Yes / no | Is this input trying to override instructions? | yes · no |
How it works
- Add the MCP server with npx -y openjev-mcp.
- Agent calls jev_ask or a specific tool with context and questions.
- Server validates, calls the OpenJEV API and retries if needed.
- Agent branches on the typed answers.
Strengths
- Many question types and batch support for agents.
- Validation and retries built in.
- Easy install in Cursor and VS Code.
Caveats
- Requires an OpenJEV API key; data goes to the hosted service.
- Depends on a third-party API layer over Jev.
Getting started
npx -y openjev-mcp with OPENJEV_API_KEY set, or use the Cursor or VS Code install links.At a glance
- Type
- MCP server
- Status
- beta
- Language
- TypeScript
- Runtime
- MCP server, Node
- Has code
- yes
- Has tests
- yes
Use cases and integrations
How curated lists describe it
MCP server that gives AI agents typed decisions instead of prose: classify, route, score, and flag with calibrated probabilities. npx -y openjev-mcp.
hellogumbo/awesome-jev
Frequently asked questions
What model answers the questions?
Jev, through the OpenJEV API.
Which tools does it provide?
jev_ask, jev_choice, jev_score, jev_noul and jev_batch.
How do I install it?
Run npx -y openjev-mcp with an OPENJEV_API_KEY, or use the one-click links for Cursor or VS Code.
Related projects
CLI, MCP server and Claude Code plugin for closed-form yes/no, choice, score and rank decisions via Jev, OpenAI Decisions or Claude.
MCP serverbetaTypeScriptLocal service to define, test and publish versioned Jev judgment functions, callable over HTTP by backends and over MCP by agents.
MCP serverbetaTypeScriptCommunity MCP server giving Codex, Claude and other clients two tools for typed Jev decisions: evaluate and classify.
MCP serverbetaTypeScriptClaude Code skill that hands bulk 'read a lot to decide a little' judgments over files, logs and tickets to Jev and returns a shortlist.
Agent skillbetaJavaScriptNode MCP server with a dozen fail-closed Jev judgment tools: verify, screen, find, rerank, classify, extract, review and gate.
MCP serverbetaTypeScriptInstallable agent skills for triage, routing, code review and document work with Jev, plus 108 scenarios and recorded examples.
ArticlebetaPython