The outer loop for coding agents. Define intent in a Plan, let the agent run, and check every pull request against it.
Human escalation for Solari browser agents — hand off the live session, solve the blocker, resume in the same session.
A plan-driven, multi-agent coding loop: separate planner, coder, and reviewer roles, each on the provider and model you choose, work together to implement your plan.
Generative UI for working with coding agents. Review, annotate, brainstorm, and give feedback through task-specific desktop workbenches.
Just Jev it. A Jev-first agent loop with reusable actions for DeepSeek Harness.