Skip to content
buildbay.

GitHub stars

211,363
—Collecting 30d data

Stats updated

210,380 stars on Sep 27 to 211,363 stars on Oct 2, up 983.6 measured star snapshots

What is OpenCode?

OpenCode focuses the coding-agent experience around a repository rather than a standalone chat. Its tools can navigate and modify code, execute development commands, use language-server information, and coordinate multiple sessions, while its provider-neutral configuration lets developers select the models and access patterns that fit their environment.

OpenCode is an open-source AI coding agent available in the terminal, as a desktop application, and through editor integrations. It can inspect repositories, edit files, run commands, use language-server context, manage parallel sessions, and work with a wide choice of model providers.

Core capabilities

  • Edits files, searches code, runs commands, and uses language-server context directly inside a repository.
  • Provides separate build and read-only planning agents, reusable custom agents, and parallel project sessions.
  • Works across terminal, desktop, and editor surfaces with a broad range of model providers.

A coding agent built around the repository

OpenCode works where development already happens. It can search files, understand symbols through language servers, make coordinated edits, execute shell commands, and show the resulting changes in a terminal interface designed for long-running repository work.

  • terminal
  • LSP
  • file editing
  • command execution

Agents and models without one fixed surface

Built-in build and plan agents separate implementation from read-only exploration, while custom agents and parallel sessions support more specialized workflows. The same project can be used from the terminal, desktop app, or editor integrations with different model providers.

  • plan agent
  • custom agents
  • parallel sessions
  • multiple providers

Where it fits

Use cases

  1. 01

    Repository development from the terminal

    Ask OpenCode to investigate a codebase, implement changes, run commands and tests, and explain the resulting diff without leaving the terminal.

  2. 02

    Planning and codebase exploration

    Use the read-only plan agent to search unfamiliar repositories, understand architecture, and prepare an implementation approach before granting write access.

  3. 03

    Model-flexible coding workflows

    Choose among supported model providers, configure task-specific agents and commands, and continue the same coding workflow from terminal, desktop, or an editor.