Agent-Reach
One CLI giving agents read access to the internet
Persistent agent teams defined in YAML
openrig builds a network of agents out of the coding tools you already run, turning scattered Claude Code and Codex terminals into a persistent team with roles, shared context and owned work. It is for developers juggling several agent sessions on a serious project who need a chain of command rather than a pile of windows.
A harness wraps a model and a rig wraps your harnesses; each seat gets a runtime, model, context and state, and the team persists between sessions instead of dying with the terminal that started it. Teams are declared in YAML, can mix runtimes such as Claude Code and Codex with owner and checker roles, and the lead agent coordinates specialists and only escalates decisions that need you; installation is via npm and requires Node 22 plus tmux on macOS or Linux.
This week’s repos: PS5 game on a PC! Turn agents into employees. AI video maker.
Top 10 Github Repos That Blew Up This Week (Oct 4, 2026)
One CLI giving agents read access to the internet
Multi-agent coding teams with roles, models and memory
Make desktop apps accessible to AI agents via CLI
Text-based browser automation CLI for AI agents
A team of AI bots living inside a chat app, each with a computer
Apache-backed agent workspace with a full audit trail