Shared Credit Tool

Shared Credit Tool

Local MCP server enabling Codex and ChatGPT to read/write files, execute commands, manage processes, use Git, and inspect images on the user's machine with full privileges.

Category
访问服务器

README

Shared Credit Tool

English | 简体中文

Shared Credit Tool is a local MCP server for two clients only:

  • Codex through newline-delimited stdio.
  • ChatGPT on the web through authenticated Streamable HTTP and OAuth 2.1.

It lets those clients read and write files, create text documents, edit code, run local applications and shell commands, manage long-running processes, use Git, and inspect images.

Full-control default: commands run with the privileges and environment of the process that starts the server. On Windows, use the included elevated launcher and approve the UAC prompt to obtain Administrator access. Anyone who can authorize against the public MCP URL can operate the computer with those privileges.

Install

py -m pip install -e ".[image]"

Python 3.11 or newer is required.

Codex

Add this to ~/.codex/config.toml:

[mcp_servers.shared_credit_tool]
command = "shared-credit-tool"
args = ["--stdio"]

The default direct-file root is the current user's home directory. The default cwd is the directory from which the server starts. Override the root with --workspace C:\ or another explicit directory when needed. Codex inherits Administrator rights only when Codex itself was started elevated.

ChatGPT Web

For the complete setup and project workflow, see the Chinese step-by-step guide.

The launcher automatically downloads the latest official Windows amd64 cloudflared release into ~/.shared-credit-tool/bin when it is missing. You can still install it system-wide first if preferred:

winget install Cloudflare.cloudflared

Then run from an Administrator PowerShell, or let the script request elevation:

powershell -ExecutionPolicy Bypass -File .\scripts\start-chatgpt.ps1

The script starts the MCP server on loopback, enables OAuth, prints the OAuth password, and opens a Cloudflare HTTPS tunnel. It detects the tunnel hostname, prints the final MCP 服务器 URL with /mcp already appended, prints 身份验证: OAuth, and copies the URL to the clipboard. In ChatGPT, enable Developer mode under Settings -> Security and login, create an app from the plus button at ChatGPT Plugins, choose OAuth, and enter the printed password on the authorization page.

Dynamic registration accepts both authorization_code and refresh_token. Access-token responses include a rotating refresh token so ChatGPT can renew a session without registering and authorizing again.

HTTP never binds publicly by default. OAuth is enabled automatically whenever HTTP starts without an explicit bearer token or SHARED_CREDIT_TOOL_AUTH_MODE=noauth.

Core Tools

The MCP catalog contains server/cwd inspection, read_file, write_file, list_dir, list_files, search_text, apply_patch, exec_command, command session polling and termination, Git status/diff/log/show/blame, and view_image.

write_file creates parent directories and writes UTF-8 text, Markdown, CSV, JSON, HTML, or source code. For DOCX, XLSX, PPTX, PDF, Office automation, or other binary formats, use exec_command with software installed on the local computer.

Defaults

  • Permission mode: dangerous (all server command gates disabled).
  • Shell environment: all (the host environment is inherited unchanged).
  • File-tool root: the current user's home directory.
  • HTTP bind: 127.0.0.1:8765.
  • HTTP authentication: OAuth 2.1 with PKCE and dynamic registration.

Use --permission-mode safe or trusted only when you intentionally want the older restrictions. See client configuration, remote access, and the tool reference in Core Tools.

Development

py -m pip install -e ".[dev,image]"
py -m unittest discover -s tests -p "test_*.py"
py -m ruff check shared_credit_tool tests

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