One first-party Skill
Install the Jev Agent Skill and use its guidance for routing, guardrails, verification, and completion checks.
Jev Agent Skill
Jev AI Agent Skill for bounded decisions
Use Jev AI Agent Skill and API together so an agent can ask a typed question while your application keeps execution, permissions, and human approval in control.
Agent decision loopSKILL + API
AGENT STATE
A coding agent is about to run a tool after inspecting a production issue.
contextPrepare the shared state
judgmentAsk one typed question
actionKeep approval in your app
Jev supplies a signal. The agent and your permissions still control execution.
- First-party guidance
- 1 Skill
- Control signals
- Typed
- Approvals
- Explicit
- Tool permissions
- Yours
Install once, use anywhere
Give an agent a decision layer it can explain
The setup is intentionally small: install the Skill, provide a server-side key, and let the agent ask a typed question when a workflow reaches an uncertain branch.
Install and configure
Use the Jev Agent Skill with your own Jev API key
npx skills add jev-ai/jev-agent-skill
export JEV_API_KEY="sk_your_key_here"
export JEV_LANGUAGE="en-US"
# Ask the agent to make a bounded decision.
# The agent can call Jev; your application still
# validates permissions and executes the action.Judgment stays separate
Keep the agent useful without making it the authority
The Skill teaches a compatible coding agent how to prepare state, choose a question type, call Jev, and interpret the result. It does not grant access to tools or decide whether your system may execute an action.
API key stays yours
Configure a Jev AI API key through the environment so it stays out of source code, prompts, and transcripts.
Typed control signals
Ask for a Choice, Score, or Noul result instead of relying on a free-form paragraph to drive control flow.
Approval remains explicit
Use Jev's probability as a signal, then apply deterministic policies and human review before high-impact actions.
Works with your agent
Add a decision layer to Codex, Claude Code, Cursor, or another compatible coding agent without rebuilding its tool loop.
Human approval boundary
Keep deletion, payment, deployment, and other high-impact actions behind the permissions and review rules you already own.
A safer agent loop
Add judgment before you add more autonomy
The Skill gives an agent a consistent way to ask Jev for a bounded signal. Your orchestrator remains responsible for tools, policy, and the final side effect.
01
Install the Skill
Add the public Jev Agent Skill to the coding agent you already use for the task.
02
Configure the key
Set JEV_API_KEY in the runtime environment and select the onboarding language when needed.
03
Connect the result
Let your orchestrator use the answer to route or pause, while tools and permissions remain behind your own application boundary.
Jev Agent Skill questions
Know what the Agent Skill does and does not do
What is the Jev AI Agent Skill?
It is a set of instructions that helps compatible coding agents prepare state, ask Jev a bounded typed question, and interpret the result for routing, guardrails, verification, or completion checks.
Does the Agent Skill execute tools or grant permissions?
No. The Skill does not create tools, grant authority, intercept shell calls, or replace your application's permission system. Your agent and application remain responsible for execution.
Which agents can use the Jev Agent Skill?
The Skill is intended for compatible coding agents such as Codex, Claude Code, Cursor, and other environments that support the skills installation workflow.
Jev AI platform
Add one decision before adding more autonomy
Let Jev judge. Let your system act.
The Jev Agent Skill does not create tools, grant authority, intercept shell calls, or replace your permissions, deterministic rules, and human approval boundaries.