OpenCode Advisor MCP

OpenCode Advisor MCP

An unofficial MCP server that lets Codex ask a read-only OpenCode advisor for a second review pass on Git changes.

Category
访问服务器

README

OpenCode Advisor MCP

OpenCode Advisor MCP is an unofficial local MCP server that lets Codex ask a read-only OpenCode advisor for a second review pass on Git changes.

This project is for people who already use Codex and OpenCode locally and want a second review tool in the loop without giving that reviewer write access.

Project Status

  • Latest tagged GitHub release: v0.2.0
  • main includes unreleased stabilization changes after v0.2.0
  • Current scope: one MCP tool, ask_opencode_advisor
  • Supported mode: source/GitHub install with a local OpenCode agent template
  • npm package publication is planned for a future release and is not available yet

Unofficial Compatibility Notice

This is an unofficial community project. It is not affiliated with, endorsed by, or sponsored by OpenAI, Codex, OpenCode, or the maintainers of those products. "Codex" and "OpenCode" are referenced only for compatibility and integration context.

Responsible Use

Use this tool only on repositories and diffs you are authorized to inspect and disclose.

Important boundaries:

  • The tool sends Git status, Git diff context, your question, and working-directory context to your configured OpenCode runtime.
  • Depending on your OpenCode configuration, that runtime may use a remote model provider. Do not assume this means "nothing ever leaves the machine."
  • The included advisor template blocks writes and denies .env reads, but that is not a complete confidentiality guarantee.
  • Keep OPENCODE_ADVISOR_ALLOWED_ROOTS narrow. Do not point it at broad parent directories unless you deliberately want that scope.

Who It Is For

  • People already running Codex locally
  • People already running OpenCode locally
  • Repositories where a second review pass is useful
  • Teams that want a read-only reviewer template with explicit install steps

What It Is Not For

  • Reviewing code you are not allowed to disclose
  • Handling secrets or sensitive repositories by default
  • Acting as an autonomous coding agent
  • Replacing your normal review process or judgment

Install From Source

Clone the repository, then install and verify dependencies:

git clone https://github.com/henrydontbbai/opencode-advisor-mcp.git
cd opencode-advisor-mcp
npm install
npm run smoke
npm test

Copy the bundled advisor template into your OpenCode agents directory:

New-Item -ItemType Directory -Force -Path <agent-dir>
Copy-Item -LiteralPath ".\agents\codex-advisor.md" -Destination "<agent-dir>\codex-advisor.md" -Force

Add this MCP config to Codex:

[mcp_servers.opencode_advisor]
command = "node"
args = ["<repo-root>\\src\\server.mjs"]
startup_timeout_sec = 30
tool_timeout_sec = 180

[mcp_servers.opencode_advisor.env]
OPENCODE_ADVISOR_ALLOWED_ROOTS = "<allowed-root-or-semicolon-list>"
OPENCODE_ADVISOR_TIMEOUT_MS = "120000"
OPENCODE_ADVISOR_MAX_DIFF_CHARS = "60000"

Replace <repo-root> with the absolute path to this source checkout.

npm Package Status

The package metadata and CLI entrypoints are present for packaging checks, but opencode-advisor-mcp has not been published to npm yet. Use the source install path above for now.

Usage

Codex gets one MCP tool:

ask_opencode_advisor

Typical prompt:

Ask opencode_advisor to review the current changes.
Focus on risks, missing tests, privacy issues, and release readiness.

The server returns structured JSON with stable error codes such as invalid_cwd, invalid_paths, git_failed, opencode_not_found, opencode_failed, and timeout.

Local Verification

From the repository:

npm install
npm run smoke
npm test

Release and acceptance steps live in:

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选