
agent-desktop
Reliable desktop computer use for any agent.
32 folders.

Reliable desktop computer use for any agent.

Persistent memory for AI coding agents.

ai CLI: text, image, video and audio generation for any agent through AI Gateway.

Codebase search powered by Jev.

Fast, cheap web agent.

Lets agents use your real logged-in browser.

Stops coding agents from claiming done without evidence.

Claude Code skills for SEO audits, ads and marketing.

Open-source computer-use drivers, sandboxes and benchmarks.

Claude Code plugin that replaces compaction summaries with Jev decisions.

Connects TypeSafe Jev to Grok Bot as a cheap decision layer.

Per-turn model and reasoning routing for Codex, driven by Jev.

Camera-only autonomous drone in MuJoCo with a small judgment model.

Typed Jev judgments as MCP tools.

Staged code-review workflow and local dashboard.

One AI trade decision every Monad block.

Generative UI framework: LLM JSON to React.

Describe a startup idea; Jev says kill, fix or ship.

Code from LangChain's Jev harness post.

Jev navigating a Neo4j graph with a classifier.

1v1 quickscope arena with Three.js and Jev.

Open coding-agent harness that runs any model.

Code from the OpenRouter Fusion announcement.

Autonomous LP rebalancing agent.

MCP server that lets a coding agent search and read documentation for APIs, libraries and services without dumping huge pages into context.

Typed semantic decisions for Unix pipes and CI.

A separate macOS VM for computer-use agents, viewed and driven from the terminal.

Code blocks from the TypeSafe/Jev docs.

Agent skills for TypeSafe's System One API.

An agent that plays Super Mario Bros. from emulator state.

System One MCP connector for fast, cheap evaluations.

A context sieve for Claude Code that judges each tool result.