Jev Agent MCP Server

About Jev Agent

Before an agent deletes a file, sends an email or says "done", it can ask Jev AI first. Jev Agent's remote MCP server (Streamable HTTP at https://jev-agent.com/api/mcp) exposes seven tools backed by the Jev model API: judge_yes_no, judge_choice (2 to 50 options), judge_score (2 to 10 levels) and compare_versions, plus graders such as grade_essay and check_resume. Each returns probabilities, so your code decides what counts as "sure enough".

Guard patterns are documented for the moments that matter: before irreversible actions, after tool calls, before claiming done, for tool choice, and before a reply. Config is documented for Claude Code, Claude Desktop, Cursor and VS Code, with an mcp-remote bridge for local-only clients.

Auth and pricing: listing tools needs no key; calls use a Jev Agent key (email-code sign-in, no card, no waitlist). Every account gets 5 free credits a month, failed requests cost nothing, and an X-TypeSafe-Key header runs Jev on your own TypeSafe account for 0 credits. The same key also works on the REST endpoints /api/v1/systemone, /api/v1/classify (8 rulesets) and /api/v1/decisions (OpenAI Decisions format, 13 models).

Made by Waller. Jev Agent is an independent site and is not affiliated with TypeSafe AI.

How to connect Jev Agent to Claude, Cursor or VS Code

Jev Agent runs as a hosted remote MCP server at https://jev-agent.com/api/mcp. Add that URL to your MCP client's config (Claude Code, Cursor, Windsurf or VS Code) and the tools appear immediately, no install required.

{
  "mcpServers": {
    "jev-agent": {
      "url": "https://jev-agent.com/api/mcp"
    }
  }
}

Claude Code: claude mcp add --transport http jev-agent https://jev-agent.com/api/mcp. Servers that require OAuth or an API key will prompt on first use.