AI-native software development

Rysh turns the terminal you already work in into an AI pair. Every pane is both a real shell and an autonomous agent with 40+ tools — so the model works where your code, your commands and your context already live.

The problem

Most AI coding tools live in a browser tab or an editor sidebar, disconnected from the shell where builds, tests, git and deploys actually run. You copy output back and forth, and the model never really sees your terminal state.

How Rysh helps

In Rysh the agent is in the pane. It reads and searches your files, edits them, runs bash, drives git, builds, runs tests and lints — each sensitive action gated by a human-in-the-loop approval so nothing runs without your say-so. Because it shares the pane’s PTY it sees real output, exit codes and errors, then iterates.

For larger work, the softdev pipeline chains phases — planning, development, linting, unit and integration testing, deployment — as events across panes, and cross-pane coordination lets one agent hand context to another.

Key capabilities

• Agent per pane with 40+ tools (read/search, edit, bash, git, build, test, lint, symbol search)
• Human-in-the-loop tool approval (approve once / always / reject with reason)
• Background bash for long-running commands, plus web search & fetch
• Multi-phase softdev pipelines and ##> events
• Project memory: notes, todos, and conversation history per workspace
• Cross-pane coordination — inspect, send, hop, and route agent output

See it in action

Get started

Every pane is a shell and an AI agent. Try it — install takes one command.