opencode subagents

Name: opencode subagents
Author: sebstaq

作者：sebstaq版本 0.1.4i★ –i在你的电脑上运行

注册名称io.github.sebstaq/opencode-subagent-mcp

安装

概览

作者提供的说明（英文原文）

Run Claude Code subagents on cheaper opencode models (DeepSeek, opencode Go) with native parity.

PopNx MCP 的三大支柱

这三项永远不会合并成一个分数：意见、测量和人气是不同的东西。

用户评价i暂无评价评价功能将在后续阶段与账户功能一起开放
我们的基础检查i1 项中通过 1 项依据公开信息检查；尚未扫描代码
使用情况i尚未统计★ –

它能帮你做什么

我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。

连接方式: 在你的电脑上运行
传输方式: stdio
分类: 其他

安装

选择你使用的 AI 应用，然后复制设置。

安装之前

基础检查：1 项中通过 1 项
在你的电脑上运行，拥有你的用户账户的权限（文件、网络、应用）。
注册中心没有列出需要填写的密钥。

你需要

npm已安装 Node.js（用于 npx 命令）

你用的是哪个应用？

ClaudeClaude CodeCursorVS CodeChatGPT

在上方选择你的应用。我们会记住它，下次按钮会显示“安装到 …”。

打开 Claude Desktop → Settings → Developer → Edit Config，把下面的内容粘贴到 claude_desktop_config.json

{
  "mcpServers": {
    "opencode-subagent": {
      "command": "npx",
      "args": [
        "-y",
        "opencode-subagent-mcp"
      ]
    }
  }
}

在终端中运行这条命令

claude mcp add opencode-subagent -- npx -y opencode-subagent-mcp

在 Cursor 中打开

或者自己复制设置

粘贴到项目中的 .cursor/mcp.json，或 ~/.cursor/mcp.json

{
  "mcpServers": {
    "opencode-subagent": {
      "command": "npx",
      "args": [
        "-y",
        "opencode-subagent-mcp"
      ]
    }
  }
}

浏览器会先询问是否打开应用，然后应用会显示设置并请你确认。密钥的值会留空，由你自己填写。

在 VS Code 中打开

或者自己复制设置

粘贴到项目中的 .vscode/mcp.json

{
  "servers": {
    "opencode-subagent": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "opencode-subagent-mcp"
      ]
    }
  }
}

浏览器会先询问是否打开应用，然后应用会显示设置并请你确认。密钥的值会留空，由你自己填写。

这个服务器在你的电脑上运行，而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。

你的选择通过 Cookie 保存在这台设备上，无需账户。

这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同，请以 README 为准。

它能做什么

它在你的电脑上运行；安装后，工具列表会显示在你的 AI 应用中。

基础检查i

我们检查的是任何人都能看到的信息；我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月6日

✓ 通过
作者与仓库一致
注册名称已通过 GitHub 账户 sebstaq 验证，该账户拥有此仓库
… 尚未检查
仍在维护
我们还没有从 GitHub 读取这个仓库
… 尚未检查
有许可证
我们还没有从 GitHub 读取这个仓库
– 不适用
端点有响应
它在你的电脑上运行，没有互联网端点
– 不适用
使用 OAuth 登录
在你的电脑上运行，改用本地设置
… 尚未检查
工具描述中没有隐藏指令
它在你的电脑上运行；我们只能读取互联网端点的工具列表

与“其他”中的其他服务器比较i

星标数多于此分类中其他 32466 个服务器的 0%
发布到注册中心的时间比此分类中其他 32466 个服务器的 92% 更新
基础检查结果等于或优于此分类中其他 32466 个服务器的 100%

与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时，星标数才计入。

常见问题

opencode subagents 安全吗？

它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前，请先看看这个服务器要求哪些权限，并使用只具备所需权限的密钥。

它需要哪些密钥或代码？

注册中心没有列出必需的密钥，不过有些服务器会在首次使用时要求你登录。

它在我的电脑上运行还是在云端？

它在你的电脑上运行。你的 AI 应用会在需要时启动它，数据不会经过作者的服务器（除非它自己调用在线服务）。

它还在维护吗？

GitHub 仓库最后更新于–，最新版本（0.1.4）于3天前进入注册中心。

哪些应用可以使用它？

支持 MCP 的应用，例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。

有什么可以替代它？

同一分类中有：Spec Driven Development, Remote Desktop Commander, HotCodePush

替代选择

同一分类中其他作者的服务器，按星标数排序。

如何看卡片：★ GitHub 星标i三大支柱：评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i

Spec Driven Development

作者：formulahendry

MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.

在你的电脑上运行★ 440MIT14个月前更新

安装

Remote Desktop Commander

作者：desktop-commander

Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.

在云端运行★ 211昨天更新

安装

HotCodePush

作者：hotcodepush-team

Manage over-the-air (OTA) updates for React Native, Expo, Capacitor and Cordova apps.

在云端运行今天更新

安装

作者的 README

打开 README

按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。

opencode-subagent-mcp

Run Claude Code subagents on cheaper opencode models — for example opencode-go/deepseek-v4.1-flash (DeepSeek) from an opencode Go subscription — while Claude Code itself keeps talking to Anthropic directly. Delegate searches, refactors, bug fixes, tests and research to a model that costs a fraction of Opus or Sonnet, and keep the tools you already use: background agents, resume via send_message, worktree isolation, .claude/agents definitions, permission prompts and structured output.

It is a stdio MCP server. Your main Claude Code session is not proxied or routed anywhere; the server only runs the subagents you delegate to it, on a private opencode serve instance it starts itself.

Install

Requires Node 22+, git, and the opencode CLI logged in to the providers you want (opencode auth login).

As a Claude Code plugin (also adds a skill that tells Claude when to delegate to opencode):

claude plugin marketplace add sebstaq/opencode-subagent-mcp

claude plugin install opencode-subagent-mcp@sebstaq-opencode

Or as a plain MCP server:

claude mcp add -s user opencode -- npx -y opencode-subagent-mcp@0.1.4

What it runs and sends

It starts a private opencode serve process on a random localhost port with a random password, and stops it when Claude Code exits. The password is generated for each run and only used between this server and that local process. The process inherits your environment so opencode can use the provider logins and API keys you already configured for it; this server never reads, stores or sends those credentials itself.
The prompts you delegate, and the files and command output the subagent reads, go to the model provider you choose through opencode (by default opencode Go), under that provider's data retention terms. opencode keeps session history in its local database on your machine. Nothing else is sent anywhere; the server has no telemetry and makes no other network calls.
It reads ~/.claude/settings.json, the project's .claude/settings.json and .claude/settings.local.json, and .claude/agents/*.md (user and project) to mirror your permission rules and agent definitions, and ~/.config/opencode-subagent-mcp/config.json if present.
With isolation: "worktree" it creates git worktrees under the project's .claude/worktrees/ and removes them when they have no changes.

This is an independent project, not affiliated with Anthropic or the opencode team.

Tools

Tool	Native equivalent	What it does
`agent`	Agent tool	Run a subagent; returns its final report, usage and worktree outcome
`send_message`	SendMessage	Continue a finished agent with its full history; also after a restart
`wait`	background completion notification	Collect a `run_in_background` agent (optionally with a timeout)
`stop`	TaskStop	Abort a running agent and return what it produced
`list`	/agents	This session's agents, Claude agent definitions and opencode agents

An agent_id keeps working after the MCP server or Claude Code restarts: wait and stop return what happened to it, and send_message continues it. Agents from before a restart do not appear in list until their id is used again, but they still exist in opencode's local session store.

agent parameters: description, prompt, subagent_type, model, isolation: "worktree", permission_mode, tools, disallowed_tools, max_turns, effort, output_schema, run_in_background, cwd.

Parity with native subagents

Native behaviour	Here
Fresh context, only the final message comes back	Same. Long output is truncated (`maxResultChars`) with the full text written to a file
Runs in the project directory	Same (cwd from MCP roots, then `CLAUDE_PROJECT_DIR`, then the server's cwd)
`isolation: worktree`	Same layout (`.claude/worktrees/` , branch `worktree-` , based on the default branch). Removed when unchanged, kept with commit/file summary otherwise. Outside paths are denied
Permission prompts surface in the main session	opencode `ask` rules are shown as MCP elicitation dialogs (allow once / always / deny, with optional feedback)
Permission modes	`default`, `acceptEdits`, `auto`, `bypassPermissions`, `plan` mapped to opencode rules; your Claude `permissions.allow/ask/deny` from user, project and local settings apply
Background agents with completion notification	Claude Code moves long MCP calls to the background automatically and notifies when they finish; `run_in_background` + `wait` for explicit control
Resume with full history	`send_message` (also works for agents from earlier sessions or after a restart, since opencode persists them)
`.claude/agents/*.md` definitions	Used via `subagent_type`: prompt, `tools`, `disallowedTools`, `permissionMode`, `maxTurns`, `effort`, `isolation`. `model` applies only when it is an opencode `provider/model`
`maxTurns` returns partial output	Same; status `max_turns`, resumable
Progress in the UI	Tool calls are sent as MCP progress notifications
Subagents cannot spawn subagents or ask the user questions	opencode's `task` and `question` tools are disabled
Structured output	`output_schema` (JSON Schema) → opencode structured output, returned as `structuredContent`

Known gaps: agents do not appear in Claude Code's /tasks agent view; there is no fork-with-parent-context; the Workflow tool's agent() cannot target them; auto mode has no safety classifier (it allows everything, including paths outside the project, except rules you configured and a fixed list of destructive commands that ask: rm -r/-rf, git push --force, git reset --hard, git clean -f, git restore/git checkout --, curl | sh, sudo, chmod -R/chown -R, dd, mkfs, npm/pnpm publish; opencode matches each command in a chain separately); the parent's permission mode is not visible to MCP servers, so the default comes from config; hooks, skills preloading, memory and per-agent MCP servers are not mapped.

Steering Claude

Claude Code picks between its native Agent tool and this one on its own, so routing is unpredictable. To make opencode the default for delegated work that is cheap to verify, paste the snippet from docs/claude-md-snippet.md into your global ~/.claude/CLAUDE.md.

Configuration

Optional ~/.config/opencode-subagent-mcp/config.json (or $OPENCODE_SUBAGENT_CONFIG):

"defaultModel": "opencode-go/deepseek-v4.1-flash", "defaultPermissionMode": "auto", "pure": true, "opencodeBin": "opencode", "maxConcurrent": 8, "maxResultChars": 80000 }

OPENCODE_SUBAGENT_MODEL and OPENCODE_SUBAGENT_MODE override the first two. pure starts opencode without external plugins.

How it works

On first use the server starts opencode serve on a random loopback port with a random password, subscribes to its event stream, and drives one opencode session per agent (agent id = session id). Permission requests from opencode are answered through MCP elicitation. The opencode process is stopped with the MCP server; leftovers from crashed servers are reaped on the next start.

Development

pnpm check   # typecheck, lint, format check, unit tests

pnpm smoke [provider/model] # end-to-end run against a real model in a temp repo

License

MIT