OpenCode Hub: Versions, Free & Local Models, Permissions and Plugins
Everything you need for OpenCode (anomalyco/opencode): connecting 75+ model providers or local Ollama models, tightening its permissive default permissions with Plan mode, and browsing plugins.
Key takeaways
- OpenCode (anomalyco/opencode) is an MIT-licensed terminal TUI and desktop coding agent built around provider freedom—supporting BYOK, ChatGPT/Copilot sign-in and local models.
- Make sure you install the active `anomalyco/opencode` project (`opencode-ai` on npm), not the archived `opencode-ai/opencode` repo.
- Safety note: OpenCode allows file edits and bash commands without prompting by default. Switch to Plan mode (Tab) and set `edit`/`bash` to `ask` in `opencode.json`.
- Its ecosystem includes memory plugins (`magic-context`), config packs (`oh-my-opencode-slim`) and ACP IDE bridges.
Key facts: pricing, models, surfaces & safety
The software is free. Run it at zero cost with local Ollama models or free provider tiers, bring your own API keys, or use the optional OpenCode Zen gateway.
Supports Anthropic, OpenAI, Gemini, DeepSeek, Qwen, OpenRouter, local engines, plus existing GitHub Copilot or ChatGPT Plus/Pro subscriptions.
Full-screen terminal TUI, desktop app, VS Code/Cursor extensions, and a headless `opencode serve` HTTP server.
`edit` and `bash` default to `allow`. Use read-only Plan mode or set `"permission": {"edit": "ask", "bash": "ask"}` before working on untrusted repos.
Extension points: opencode.json, AGENTS.md, Plugins and ACP
OpenCode extensions integrate through four documented surfaces:
opencode.json
Project or global config for providers, `permission` rules, MCP servers and plugin packages.
AGENTS.md & Custom Commands
Reads standard `AGENTS.md` files and custom slash commands in `.opencode/command/`.
OpenCode Plugins
JS/TS plugin hooks that add persistent memory (e.g. magic-context), custom tools or UI widgets.
OpenCode Server & ACP
Headless server and Agent Client Protocol support for external IDEs and multi-agent orchestrators.
Setup & troubleshooting guides
OpenCode setup and first-use guide
Tell OpenCode 1 from 2, install it, connect ChatGPT, Copilot, an API key or a local model, and tighten its permissive defaults with Plan mode — with WSL notes and official sources.
Setup and first-use guideCC Switch guide
Install CC Switch from official sources, switch API providers for Claude Code, Codex, OpenCode and more, understand what routing mode writes to your config, and the risks of key storage and third-party relays.
Companion tool guideOpenCode ecosystem & extensions
Plugins, config packs, switchers and orchestration layers in the directory built for OpenCode, by stars; each has its own FARS review.
Multi-agent orchestration, parallel sessions & workbenches
Oh My OpenAgent
An OpenCode and Codex CLI layer for orchestrated coding workflows and integrated developer tools.
Paseo
One interface to orchestrate multiple coding agents (Claude Code, Codex, Copilot, OpenCode, Pi) from desktop and mobile.
Oh My OpenCode Slim
An OpenCode plugin that routes development work across specialized, configurable multi-model agents.
Emdash: Parallel AI Coding Agent Desktop
Run multiple coding agents in parallel, each in its own Git worktree, and review & merge in one place.
MyClaude Multi-Agent Workflow
Claude Code plans and verifies while Codex, Gemini or OpenCode backends do the actual coding.
Nimbalyst Visual Coding Workspace
A desktop workspace for running Codex and Claude Code sessions in parallel while reviewing file changes visually.
Cross-session memory & context tools
Plugins, routing & companion extensions
img2threejs — Code-Only 3D Reconstruction
Rebuild objects from a reference image as code-only, procedural, quality-gated, animation-ready Three.js models, token-efficiently.
ccusage
Analyze token usage and costs for coding-agent CLIs from local data, with daily, weekly, monthly and session reports.
Plannotator
Visually annotate your coding agent's plans and diffs in the browser and send feedback back with one click.
OpenChamber
Open-source workspace to run, supervise, and review AI coding work across desktop, browser, editor, and mobile.
HumanLayer
Run coding agents on your own machines and steer them from one place — tasks, sessions, plans, and reviews.
iPolloWork
A local-first visual AI workbench that turns one goal into editable code, documents, presentations, websites, designs, and videos—an open alternative to Codex and Claude Code.
Autoresearch: Autonomous Iteration Engine
Turn Claude Code, OpenCode, or OpenAI Codex into a relentless improvement engine: set a goal, define a metric, and let it iterate modify→verify→keep/discard automatically.
AI-DLC Workflows
One harness-neutral core, native across major AI coding agents, turning AI agents into verifiable, self-correcting engineering workflows.
OpenAgents Control (OAC)
AI agents that learn your coding patterns and generate matching code every time, with approval gates for controlled execution.
Council of High Intelligence
Put consequential decisions through independent analysis, direct challenge, and a verdict that preserves dissent.
gnhf (Good Night, Have Fun)
Tell your coding agent one objective before bed; it commits small changes all night and leaves you a clean branch with a full log.
MonoCode
A desktop UI that unifies Claude Code, Codex, Cursor and other coding agents into tabbed sessions, so you stop juggling terminal windows.
opencode.nvim
Connect Neovim to OpenCode within your existing editor workflow.
abtop Agent Session Monitor
A terminal dashboard for local AI coding sessions, tokens, context, limits, and ports.
Maestro: Agent Orchestration Command Center
Cross-platform desktop app for orchestrating multiple AI coding agents in parallel, with automation via playbooks and Git worktrees.
Deep Research Skills
A structured deep research workflow for Claude Code, OpenCode, and Codex with human-in-the-loop control at every stage.
Browse OpenCode skills on FollowSkills → Switching OpenCode between model providers? Read the CC Switch guide →
OpenCode comparisons & rankings
AI coding agents, ranked
Gemini CLI, Cline, OpenCode, Kilo Code, Aider, Qwen Code and more ranked by one FARS source review, with license, stars and maintenance; Claude Code and Codex CLI are listed but unscored.
View full category ranking →Open Source Coding Agents, Compared
Open source coding agents ranked by FARS review — score, trust, license, language and maintenance side by side, with what each one is and isn't good at.
12 reviewed agents comparedRelated concepts
Frequently asked questions
How do I choose between OpenCode and Claude Code?
Pick Claude Code if you want Anthropic's first-party harness with conservative permission prompts out of the box. Pick OpenCode if you want an MIT-licensed tool that mixes DeepSeek, local Ollama, OpenRouter or Copilot subscriptions—just tighten its permissions first.
How do I make OpenCode ask before running bash commands?
Add `"permission": {"edit": "ask", "bash": "ask"}` to `opencode.json`, or press Tab inside the TUI to switch to read-only Plan mode.
What is OpenCode Zen?
OpenCode Zen is an optional hosted model gateway curated by the OpenCode team. It is not required—you can connect your own provider keys or local models directly.