CCConfig MCP server

Commit Nov 12, 2026

Manages configurations, API keys, MCP servers, and skills for multiple AI coding CLI agents through a unified interface.

Also known as ShaneLevs/CCConfig

Developer Tools

GitHub ↗

Stars: 3
Tools: —
Downloads, 30 days: —
Latest version: —
Last commit: Nov 12, 2026
License: MIT

What is CCConfig MCP?

CCConfig manages configurations, API keys, MCP servers, and skills for multiple AI coding CLI agents through a unified interface.

Summary generated from the project's README

About CCConfig MCP server

Developers using multiple AI coding CLI agents can centralize their entire workflow through this configuration management tool. It provides a unified interface to manage API keys, models, MCP servers, and skills for fifteen different applications including Claude Code, OpenCode, Pi Agent, and Cline. Instead of manually editing TOML, YAML, or JSON files, users can switch providers and models with a single click.

The system operates as a local desktop application for macOS and Windows or as a uTools plugin. It features a local gateway that exposes selected models via HTTP to any local agent, supporting protocol conversion for Anthropic, OpenAI Chat, and OpenAI Responses. It also includes a model mapping mode that allows up to three upstream models to be assigned to a single alias, utilizing strategies like random calling or automatic error switching. Users can monitor token usage through heatmaps and stacked bar charts, while also managing skills from SkillHub or ModelScope.

What you can do with it

1Centralized API Key ManagementSwitch between different AI model providers and API keys across multiple coding agents without manual file editing.
2Unified MCP Server ControlManage MCP server configurations and discover available tools through a real-time canvas.
3Protocol Translation via Local GatewayExpose a specific model to any local agent by converting between different communication protocols.
4Token Usage MonitoringTrack how much you are spending on different models and agents using visual charts.

Written from the project's README. Every use case is checked against a quote from it.

Configuration

CCCONFIG_NODE

Name	Description
CCCONFIG_NODE	—

Automated checks

Ran Nov 14, 2026Full report →

58/100

Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points

Show this score in your README

Source repository linked and reachableshanelevs/ccconfig is public and reachablegithub.com/shanelevs/ccconfig ↗Pass
Package names this repository as its sourceNot checked
Build provenance (Sigstore attestation)Not checked
Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
Commit or release in the last yearLast commit or release 2026-11-12github.com/shanelevs/ccconfig ↗Pass
OpenSSF ScorecardNot checked
Package name not close to a popular packageNot checked
No critical advisory in the package or its dependenciesNot checked
No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
Destructive tools are labelledNot checked
Remote endpoint answers the MCP handshakeNot checked

README

View on GitHub

CCConfig

简体中文 | English

多应用 AI 配置管理工具，同时提供 uTools 插件与独立桌面应用（macOS / Windows）两种形态。

把十五个 AI 编程 CLI 的供应商、模型、MCP、Skill、插件与 Token 用量收进一个界面，一键切换供应商，不必再手改 TOML/YAML/JSON。

功能特性

统一应用切换 — 十五个 AI CLI + 跨 agent 的「通用配置」，各自独立配置，一键切换
API 配置管理 — 逐应用读取、保存、切换供应商配置，本工具不管理的字段原样保留（未知键往返不丢）
供应商/模型主数据 — 共享的供应商与模型库（API Key AES-256-CBC 加密），支持 OpenAI Chat Completions / OpenAI Responses / Anthropic Messages / Google Generative AI 四类协议，一键下发到任意受支持的 agent
本地网关 — 把选中的供应商/模型暴露在 http://127.0.0.1:<port>，对本机任意 agent 提供 Anthropic / OpenAI Chat / OpenAI Responses 三种协议，跨协议自动转换（含流式）；「模型映射」模式下用固定映射名（如 pro / plus）对外，每个映射名可配置最多 3 个上游模型，支持只使用第一个 / 报错自动切换 / 随机调用 / 依次调用四种调配策略；换上游只改映射，响应里的 model 也会改写成映射名（默认开，可关），所以各 agent 的使用统计显示的是映射名而不是上游真实模型名
MCP 管理 — 逐应用 MCP Server 配置 + 实时工具发现画布；通用配置里的共享 MCP 按机器开关镜像写入本地 JSON 文件
Skill 与插件管理 — 从 SkillHub / 魔搭社区安装 Skill，或粘贴 skills.sh 详情页链接 / npx skills add 命令一键安装到 ~/.agents/skills，并用 skills 同级的 .disabled 停用目录启停；管理 Claude Marketplace 插件、OpenCode plugin 数组、Pi Extension
使用统计 — Token 用量、模型分布、GitHub 风格贡献墙热力图；通用统计页另有按天/按周堆叠的 Token 柱状图，可按模型或 Agent 拆分看用量构成
订阅用量 — 手动添加已订阅的服务（目前 open-design），用本机浏览器的登录态代查 5 小时／周额度，同时显示在桌面版托盘面板
批量编辑 / 导入导出 / 外观 — 按供应商聚合一键改 URL+Key，JSON 文件或压缩加密字符串导入导出；外观支持跟随系统 / 固定浅色 / 固定深色，深色下可选动态背景特效
中英双语 — 默认跟随系统语言，可在「设置 → 通用」中切换简体中文 / English

支持的应用

应用	配置文件
Claude Code	`~/.claude/settings.json`（MCP 在 `~/.claude.json`）
OpenCode	`~/.config/opencode.json` / `opencode.jsonc`
Pi Agent	`~/.pi/agent/settings.json` + `models.json`
omp	`~/.omp/agent/models.yml` + `config.yml`（统计读 `sessions/` 与 `archive/sessions/`）
Reasonix	`~/.reasonix/config.toml` + `.env`（统计读 `stats/*/.jsonl`）
Codex	`~/.codex/config.toml` + `models.json`
Kimi Code	`~/.kimi-code/config.toml`
MiniMax Code	`~/.minimax/config.yaml`
Qoder	`~/.qoder/settings.json` / `~/.qoder-cn/settings.json`
ZCode	`~/.zcode/v2/provider_config.json`
Hermes	`~/.hermes/config.yaml`
DSH（DeepSeek Harness）	`<DSH_HOME>/profiles/<profile>/cordis.patch.yml`
Grok Build（xAI 官方 CLI）	`~/.grok/config.toml`
Cline	`~/.cline/data/settings/providers.json` + `models.json`（MCP 在 `cline_mcp_settings.json`）
Command Code	`~/.commandcode/providers.json` + `settings.json`（key 在 `auth.json`，MCP 在 `mcp.json`，Taste 在 `taste/`）

各应用的字段级规则见 docs/architecture.md。

安装

uTools 插件 — 在 uTools 插件商店搜索 CCConfig 安装。

桌面应用（macOS / Windows） — 从 Releases 下载 .dmg（macOS）或 *-setup.exe（Windows）。发布 Release 时 GitHub Actions 会自动构建并挂上这两个安装包；都需要本机已安装 Node.js ≥ 18（应用刻意不内置运行时）。

本地构建（当前平台）：

npm install
npm run tauri:build
# macOS → release/CCConfig_<版本>_<arch>.dmg
# Windows → release/CCConfig_<版本>_x64-setup.exe
# Tauri 原始产物仍在 src-tauri/target/release/bundle/ 下，
# scripts/collect-release.mjs 在构建末尾复制一份到仓库根的 release/

应用数据存放于 ~/.ccconfig/（Windows 为 %USERPROFILE%\.ccconfig）。关闭窗口只是隐藏到托盘，本地网关继续运行：左键单击托盘图标打开状态面板（网关统计、订阅用量、快捷动作），右键单击显示窗口。

如果 Node 是通过 nvm/fnm/volta 安装的、从 Finder / 开始菜单启动后应用找不到它，可显式指定：

# macOS
CCCONFIG_NODE=$(which node) open -a CCConfig
# Windows（PowerShell，先设置变量再启动应用）
$env:CCCONFIG_NODE = (Get-Command node).Source
& "C:\Path\To\CCConfig.exe"

使用方式

uTools 中输入关键词唤起对应应用：CCConfig / 通用配置（指向通用配置）、Claude Code配置、OpenCode配置、Pi Agents配置、omp配置、Reasonix配置、Codex配置、Kimi Code配置、MiniMax Code配置、dsh配置。粘贴 SkillHub / 魔搭链接、skills.sh 详情页链接或 npx skills add 命令进入 Skill 安装；pi install <包名> 进入 Pi Extension 安装。

独立桌面应用没有搜索框：启动后进入「通用配置」，通过左侧栏切换应用。

开发

npm install
npm run dev          # uTools 插件开发服务器 (localhost:5173)
npm run build        # uTools 插件 → dist/
npm run build:app    # 桌面版渲染层 → dist-app/ + sidecar 资源
npm run tauri:dev    # 桌面版开发模式
npm run tauri:build  # 桌面版打包 → release/*.dmg（macOS）或 *-setup.exe（Windows）

两个发行版共用 src/ 与 public/preload/，只有外壳不同。架构说明与维持这一点的约束见 AGENTS.md 与 docs/architecture.md。

发版（维护者）

npm run bump -- X.Y.Z   # 同步 5 处版本号，勿手改单个文件
npm run version:check
npm run i18n:check
npm run routes:check
npm test
npm run build
# 提交本次发布改动并推送 main，然后：
git tag vX.Y.Z && git push origin vX.Y.Z
gh release create vX.Y.Z --generate-notes

tag 必须等于 v + package.json 版本；发布正式 Release（非 draft / prerelease），才会触发桌面版构建并进入 releases/latest。
桌面版自动更新依赖 Actions Secret TAURI_SIGNING_PRIVATE_KEY，私钥必须与 src-tauri/tauri.conf.json 的 plugins.updater.pubkey 成对；私钥有密码时再加 TAURI_SIGNING_PRIVATE_KEY_PASSWORD。Secret 缺失时只发布普通安装包，不会生成更新包。
CI 的 updater-manifest job 会合并各平台 .sig 并上传 latest.json。发布后确认 Release 包含 dmg / exe / .app.tar.gz / .sig / latest.json。
更新源：https://github.com/ShaneLevs/CCConfig/releases/latest/download/latest.json。版本号只增不减，不要复用或覆盖同一版本的 tag / 资产；发坏了升补丁版本重发。
macOS 更新包需要 app 构建目标（workflow 已用 --bundles app,dmg）；uTools 插件仍需手动上传 dist/。

技术栈

渲染层 Vue 3 + Vite + TDesign Vue Next；文案用 vue-i18n 与共享的 shared/i18n JSON（Vue / preload / Rust 三端共用）。文件读写与网络请求全部在打包后的 preload 里，经 uTools API 或 Tauri v2 Node sidecar 暴露。配置解析用 json5 / js-yaml / smol-toml；MCP 工具发现用 @modelcontextprotocol/sdk；背景特效用 ogl。

License

MIT

Read full README

CCConfig MCP questions

How are my API keys secured?

API keys are protected using AES-256-CBC encryption within the shared provider and model library.

Can I use CCConfig to host a local gateway for my models?

Yes, it can expose selected providers or models at a local HTTP port, providing Anthropic, OpenAI Chat, or OpenAI Responses protocols with automatic cross-protocol conversion.

Does CCConfig manage MCP servers?

Yes, it allows for per-application MCP server configuration and features a real-time tool discovery canvas.

Facts

Nov 15, 2026

Stars: 3
Downloads, 30 days: —
Latest version: —
Last commit: Nov 12, 2026
License: MIT
Transport: —
Package: —

ShaneLevs ↗

Community publisher

Published by an independent maintainer.

Alternatives

Works with

Claude Code Cursor VS Code Claude Desktop Codex Gemini CLI Windsurf Devin

Badge for your README

Show your HostedMCP checks score. The badge links readers to install instructions for every client.

[![HostedMCP checks](https://api.hostedmcp.org/v1/badge/server/shanelevs/ccconfig.svg)](https://hostedmcp.org/servers/shanelevs/ccconfig/checks)

Maintainer?

Claim this listing to edit it and add the HostedMCP badge to your README.

Claim listing

Before you install

Servers run with the permissions you give them. Use the narrowest token or database role that does the job.

About CCConfig MCP server

What you can do with it

1Centralized API Key ManagementSwitch between different AI model providers and API keys across multiple coding agents without manual file editing.
2Unified MCP Server ControlManage MCP server configurations and discover available tools through a real-time canvas.
3Protocol Translation via Local GatewayExpose a specific model to any local agent by converting between different communication protocols.
4Token Usage MonitoringTrack how much you are spending on different models and agents using visual charts.

Written from the project's README. Every use case is checked against a quote from it.

Name

Description

CCCONFIG_NODE

—

应用

配置文件

Claude Code

~/.claude/settings.json（MCP 在 ~/.claude.json）

OpenCode

~/.config/opencode.json / opencode.jsonc

Pi Agent

~/.pi/agent/settings.json + models.json

omp

~/.omp/agent/models.yml + config.yml（统计读 sessions/** 与 archive/sessions/**）

Reasonix

~/.reasonix/config.toml + .env（统计读 stats/**/*.jsonl）

Codex

~/.codex/config.toml + models.json

Kimi Code

~/.kimi-code/config.toml

MiniMax Code

~/.minimax/config.yaml

Qoder

~/.qoder/settings.json / ~/.qoder-cn/settings.json

ZCode

~/.zcode/v2/provider_config.json

Hermes

~/.hermes/config.yaml

DSH（DeepSeek Harness）

<DSH_HOME>/profiles/<profile>/cordis.patch.yml

Grok Build（xAI 官方 CLI）

~/.grok/config.toml

Cline

~/.cline/data/settings/providers.json + models.json（MCP 在 cline_mcp_settings.json）

Command Code

~/.commandcode/providers.json + settings.json（key 在 auth.json，MCP 在 mcp.json，Taste 在 taste/）

npm install npm run tauri:build # macOS → release/CCConfig_<版本>_<arch>.dmg # Windows → release/CCConfig_<版本>_x64-setup.exe # Tauri 原始产物仍在 src-tauri/target/release/bundle/ 下， # scripts/collect-release.mjs 在构建末尾复制一份到仓库根的 release/

npm install npm run dev # uTools 插件开发服务器 (localhost:5173) npm run build # uTools 插件 → dist/ npm run build:app # 桌面版渲染层 → dist-app/ + sidecar 资源 npm run tauri:dev # 桌面版开发模式 npm run tauri:build # 桌面版打包 → release/*.dmg（macOS）或 *-setup.exe（Windows）

npm run bump -- X.Y.Z # 同步 5 处版本号，勿手改单个文件 npm run version:check npm run i18n:check npm run routes:check npm test npm run build # 提交本次发布改动并推送 main，然后： git tag vX.Y.Z && git push origin vX.Y.Z gh release create vX.Y.Z --generate-notes