CODING AGENT INTEGRATION

Jev Agent Skill

Configure the Jev Agent Skill for Codex, Claude Code, or Cursor, then use typed decisions for routing, tool checks, and task completion reviews.

Install the public skill

The Jev Agent Skill explains when to use Choice, Score, or Noul and how to interpret the response. Install it with the agent skill installer, then ask your coding agent to use Jev Agent.

npx skills add jev-ai/jev-agent-skill

Point requests to this API gateway

The upstream skill defaults to https://thejevai.com. Set JEV_API_BASE_URL explicitly if you want it to call Best Jev AI. Provide the API key through the environment used to launch your coding agent; never place a real key in a prompt or repository.

export JEV_API_BASE_URL="https://bestjevai.com"
export JEV_API_KEY="sk_your_key_here"
export JEV_LANGUAGE="en-US"

Replace the placeholder locally with a real key from your account settings. Do not paste the real value into chat, code examples, or source control.

Keep the decision separate from the action

Give the model the smallest useful state and one clearly bounded question. Your application or coding agent should validate the result and remain responsible for permissions and final actions.

  • Choice selects one item from an allowlisted set.
  • Score evaluates an ordered scale such as urgency or review priority.
  • Noul estimates whether a yes/no statement is true.
  • Use thresholds based on your workflow and retain deterministic checks and human approval for consequential actions.

Setup and API documentation

Open the Playground