DOCS

Jev Agent: a decision step for your workflow

Try a bounded judgment before connecting an agent. This independent workspace uses Jev through OpenRouter. Your main model still plans the work; your application controls permissions and actions.

JEV / AGENT
Remove passwords, payment details, and unnecessary personal information.

Typed result

Choose a workflow, edit the example, and run when ready.

A high score is not a guarantee. Keep a person in control of consequential actions.

Your application sends a bounded request; Jev returns a typed answer; your policy controls the next action.
Your application sends a bounded request; Jev returns a typed answer; your policy controls the next action.

Choose one decision boundary

Pick the smallest question that changes what your application should do next. Tool guard reviews a proposed action; model route compares candidates you actually have; task route chooses a work path. Research review evaluates evidence already collected. Completion review compares work with acceptance criteria. None of these tools browse, execute another tool or run your tests.

The custom option accepts the same typed questions as the REST API. Keep exact calculations, eligibility checks and authorization in application code. A model judgment must never widen the permission the user already gave.

jev_guard_tool_call → allow | confirm | review | deny
jev_route_model → one supplied candidate id
jev_route_task → proceed_fast | deep_review | split_task | block
jev_check_research → accept | verify_more | reject
jev_review_completion → complete | verify_more | incomplete
jev_decide → your choice, noul or score questions

Inspect the input before running

The editor contains synthetic examples, not private account data. Replace them with short, relevant facts. Remove passwords, keys, payment details and unnecessary personal information. Missing facts stay missing; the model cannot inspect your computer or fetch hidden evidence.

Run sends your input to the live provider. The page uses your existing browser session and needs no API key. Anonymous access shares the playground limit of three attempts per source address per UTC day, subject to shared capacity. Signed-in decisions use your workspace credits; a successful new request uses one credit.

Account and credits · Privacy and processing

Read the answer before taking action

A typed choice is a recommendation within the labels you supplied. Check the review flag and the evidence behind the answer. High confidence is not a guarantee. The tool does not approve a refund, deploy code or confirm a claim just because it selected a favorable label.

If your policy requires a person, keep that check outside the model. When facts or scope change, run a new evaluation. When the connection fails without a confirmed result, retry the unchanged input with the same request identifier.

Design human review

Connect after the small trial works

Use MCP for a compatible agent or REST for a server integration. Both use a workspace key created in your account with active paid API access. A TypeSafe, OpenRouter or another community site key is not a Jev API Pro key. This page does not collect any of those credentials.

MCP setup · REST quickstart · Workspace pricing

Try a decision ↗

Continue your workflow

DAILY BONUS

2 free credits, every day

Claim 2 free credits each UTC day — 14 across seven claims. No purchase needed. Credits do not expire.

    Resets at 00:00 UTC. No streak required.

    Try a decision →

    Take the next decision into your workspace.

    3 anonymous attempts per day · 20 signup credits · No card for the trial

    Sign in ↗

    Tell us what you need

    Describe the workflow you want to classify, the volume you expect, and the outcome you need. We reply by email.

    10–2,000 characters. Never include passwords, API keys, or payment details.