Launches on August 28, 2026

One CLI for all your robots. Connect them, command them, and let them work together, each with an LLM for a brain (Claude, OpenAI, Gemini, Grok, or local via Ollama or vLLM) and a decision LLM for multiple choice (Jev, Laya, Kev). Drives Microduck, Open Duck Mini, LeRobot, XLeRobot, AlohaMini, ToddlerBot, or any ROS 2 base from a laptop, offboard.

A curated OpenCode configuration pack with specialized agents, skills, rules, and knowledge-graph tooling.

A minimal, explicit, provider-safe harness for the Pi coding agent

Tag $agent inside your Pull Requests and Issues to let Jev decide which model and reasoning to use.

Metered, budget-governed AI Gateway usage for Convex: cost & usage tracking, per-user and per-action spend/rate limits, request logging, and re-run.

Chusky can connect apps, run shell commands, browse the web, handle real-time trigger events, and execute across major SaaS platforms from Telegram, linked channels, or a terminal.

帮助非技术背景的人，和 AI 一起把想法做成可以使用的产品。Skills for turning ideas into working products.

A live semantic map and diagnostics dashboard for long-running pi sessions

Declarative Arch/CachyOS workstation bootstrap and agent operating environment for Codex, Claude Code, and Pi, with reusable skills and development workflows.

Institutional-grade stock analysis skill: fundamentals & technicals (multi-agent hedge-fund framework)

Auto-transcribes lectures based on class schedule into formatted notes (local Whisper + Claude, optional Q&A speaker detection)

Provider-neutral language model catalog and client for Rust.

Adaptive memory routing for agent harnesses -- a router that decides which memory shards to contact by cost, confidence, and access scope, carried by a real agent harness.

Shared memory layer for coding agents across harnesses (Claude Code / Codex / Cursor / Kiro), exposed via MCP

Pi extension: Merge Dev gateway provider (OpenAI Responses API) hosting GLM 5.3 Flash.

DocSentry: A GitHub Action that monitors a codebase, detects when code changes make documentation inaccurate, identifies the specific stale sections, and either auto-generates a PR with corrected docs or flags the discrepancies for human review.