What is Jev?
Start with the model’s role, its output types and the limits of the available evidence.
Understand it.
Then build with it.
Introductions, setup paths and architecture notes for working with Jev. Start with the basics or go straight to a question in your own workflow.
Choose your starting point
Find a setup path for a coding agent, browser workflow or your own application.
Where Jev fits in an agent
Follow the decision from application state to a typed result and an action in your code.
Read the coding-agent engineering PDF
Understand the independent working note, its provenance and the context behind its claims.
ARCHITECTURE & METHODS
Find the decision in your workflow.
Explore routing, tool selection, evaluation and related Agent settings. These notes explain design choices; project examples link back to the case catalog.
Browser Agents
Which browser action should the agent take next, and has the task succeeded?
Agent Routing
Which path should handle this state?
Tool Selection
Which available tool is the best next action?
Agent Evaluation
Did the agent complete the task correctly and safely?
Guardrails
Is this action allowed, safe, and sufficiently supported?
MCP / Skills
How do we make fast typed judgment available to the agent runtime?
Search & Semantic Retrieval
Which documents, chunks, or entities match this unstructured query or indexing taxonomy?
Realtime & Audio Agents
What is the immediate reactive action or objection response based on live transcribed speech and state?
Coding Agents
Is this generated patch safe, syntactically sound, and aligned with user repo constraints before execution?