xiaobenyang-mcp-2-1
# Local Filesystem MCP Server
A Model Context Protocol (MCP) server for exploring local filesystems with glob and grep tools.
Built with [Smithery SDK](https://smithery.ai/docs)
## Features
This server provides two essential tools for filesystem exploration:
- **glob** - Find files matching patterns (e.g., `**/*.ts`, `src/**/*.js`)
- **grep** - Search for text patterns in files
## Prerequisites
- **Smithery API key**: Get yours at [smithery.ai/account/api-keys](https://smithery.ai/account/api-keys)
## Getting Started
1. Install dependencies:
```bash
npm install
```
2. Start development server:
```bash
npm run dev
```
The server will run locally and provide access to your filesystem through the MCP protocol.
## Configuration
You can customize the working directory in your `smithery.yaml` config:
```yaml
runtime: typescript
target: local
config:
workingDirectory: /path/to/your/directory
```
By default, it uses the current working directory.
## Development
Your code is organized as:
- `src/xiaobenyang_mcp_tools.ts.bak` - MCP server with glob and grep tools
- `smithery.yaml` - Runtime specification with `target: local` for filesystem access
Edit `src/xiaobenyang_mcp_tools.ts.bak` to add your own filesystem tools.
## Build
```bash
npm run build
```
Creates bundled server in `.smithery/`
## Deploy
This server uses `target: local` in `smithery.yaml`, which means it's designed to run locally with filesystem access. It cannot be deployed to remote Smithery hosting.
## Learn More
- [Smithery Docs](https://smithery.ai/docs)
- [MCP Protocol](https://modelcontextprotocol.io)
TDQS
Scored across 2 tools
The two tools have perfectly distinct purposes: get_all_dishes retrieves a list of dish names, while get_dish_content fetches detailed content for a specific dish. There is no overlap or ambiguity between these operations.
Both tools follow a consistent verb_noun pattern with 'get_' prefix and snake_case naming. The naming is predictable and clear across the set.
With only 2 tools, this server feels thin for a dish/recipe domain. While the tools cover basic retrieval, typical food-related servers would include operations like search, filter, create, update, or delete dishes for a more complete surface.
The server only provides read operations (list and get details), lacking any create, update, delete, or search capabilities. This is a significant gap that will limit agent workflows, as there's no way to modify or interact with dish data beyond viewing it.