Caboose MCP
A comprehensive personal AI toolserver that exposes 105 tools across 25 groups—including Slack, Docker, and calendar management—to Claude and VS Code. It features a Go-based MCP server integrated with a custom VS Code extension and n8n workflow automation for streamlined personal productivity.
README
caboose-mcp
Personal AI toolserver monorepo. Exposes 105 MCP tools to Claude and VS Code via a Go server, with a companion VS Code extension and n8n workflow automation.
Packages
| Package | Description |
|---|---|
packages/server |
Go MCP server — 105 tools across 25 groups (focus, slack, docker, calendar, etc.) |
packages/vscode-extension |
VS Code sidebar + status bar; connects to the server over stdio or HTTP |
packages/n8n |
Custom n8n image with pre-loaded caboose-mcp workflows |
Quick Start (Docker)
# 1. Clone
git clone https://github.com/caboose-mcp/caboose-mcp
cd caboose-mcp
# 2. Configure
cp .env.example .env
# Edit .env — at minimum set SLACK_TOKEN, GITHUB_TOKEN, GPG_KEY_ID
# 3. Start
docker compose -f docker/docker-compose.yml up -d
# MCP server: http://localhost:8080/mcp
# n8n: http://localhost:5678
The server mounts your ~/.claude directory so all data (notes, sessions, secrets, audit logs) persists on the host. n8n workflows are auto-imported on first run.
Quick Start (Local / Claude stdio)
# Build the server
cd packages/server
export PATH=$PATH:/usr/local/go/bin
go build -o caboose-mcp .
# Add to Claude's .mcp.json
{
"mcpServers": {
"caboose": {
"type": "stdio",
"command": "/path/to/packages/server/caboose-mcp"
}
}
}
VS Code Extension
cd packages/vscode-extension
pnpm install
pnpm compile
# Press F5 in VS Code to launch the extension host
Configure via Settings → Caboose MCP:
cabooseMcp.transport:http(default) orstdiocabooseMcp.host/cabooseMcp.port: point at the Docker servercabooseMcp.binaryPath: path to the built binary (stdio mode)
n8n Integration
Three workflows are pre-loaded:
- Event Receiver — receives
gate_fired,source_changed,focus_started, etc. - Daily Digest — 8am: calls
source_digest+si_tech_digest - Nightly Scan — midnight: calls
si_scan_dir+source_check
See docs/n8n.md for full documentation.
Infrastructure (Terraform)
Provisions AWS resources: IAM user (Bedrock), S3 bucket (cloudsync), ECR repo (Docker images).
cd terraform/aws
cp terraform.tfvars.example terraform.tfvars
# Edit terraform.tfvars
terraform init
terraform plan
terraform apply
See terraform/aws/README.md for details.
Development
Root scripts (pnpm)
pnpm build # compile extension + build Go server
pnpm server:build # Go build only
pnpm extension:compile # tsc compile only
pnpm extension:package # package .vsix
pnpm docker:build # docker build
pnpm docker:up # docker compose up -d
pnpm docker:down # docker compose down
pnpm docker:logs # follow compose logs
Go server
cd packages/server
go build -o caboose-mcp .
go vet ./...
No database or migration step — the server uses flat JSON files under ~/.claude/.
Adding a tool
# From Claude or the extension, call:
tool_scaffold # generates a new tools/mytool.go skeleton
tool_write # writes the file
tool_rebuild # runs go build
Or edit packages/server/tools/ directly and run go build.
Releases
| Tag pattern | Action |
|---|---|
server/v1.2.3 |
Builds Go binary for linux/amd64 + arm64; pushes Docker image to GHCR |
extension/v0.2.0 |
Packages .vsix; creates GitHub Release with the asset attached |
Repository layout
packages/
server/ Go MCP server
vscode-extension/ VS Code extension (TypeScript)
n8n/ Custom n8n image + pre-built workflow JSON
docker/
Dockerfile Multi-stage Go build → alpine runtime
docker-compose.yml server + n8n services
docs/
n8n.md n8n integration guide
terraform/
aws/ IAM, S3, ECR
.github/workflows/
ci.yml Go vet/build + TS compile on PR
docker.yml Build + push to GHCR on main / server/v* tag
extension.yml Package + release VSIX on extension/v* tag
License
MIT
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。