
Busabase
AI Knowledge Management
An open-source shared workspace and system of record for AI agents.
AI Developer Tools
Open-source terminal AI coding agent with memory, swarms, and background tasks

Explore other tools in related categories.

AI Knowledge Management
An open-source shared workspace and system of record for AI agents.

AI Assistant
Learns how you work. Then does it for you.

AI Accounting
Online Pakistani tax return filing, wealth reconciliation, and NTN services
Jcode.sh is the website for jcode, an open-source AI coding agent designed for the terminal. Written in Rust, jcode is built for speed and efficiency, with a focus on enabling massive parallelism. It features durable semantic memory, background task management, and multi-agent swarm coordination, all within a single repository. The project is MIT licensed and backed by Y Combinator.
To install jcode, run the following command in your terminal:
curl -fsSL https://jcode.sh/install | bashAlternatively, you can download pre-built binaries for macOS, Linux, or Windows from the GitHub releases page. After installation, you can start a session by running jcode in your terminal. The tool provides an interactive TUI (Terminal User Interface) for rich agent work, and also supports headless runs via jcode run for automated tasks.
jcode is open source and free to use. The website lists a pricing page, but the software itself is available at no cost, with the license being MIT.
Is jcode free? Yes, jcode is open source and free to use, with an MIT license.
What operating systems are supported? jcode supports Linux, macOS, and Windows.
How does jcode handle memory? jcode uses semantic embeddings to store and retrieve memories, allowing the agent to recall relevant information automatically.
Can jcode run multiple agents at once? Yes, jcode is designed for massive parallelism, with low memory overhead per session, making it feasible to run dozens of agents.
Does jcode support background tasks? Yes, jcode has built-in background task management, allowing long-running commands to continue while the agent works on other things.