yourname-mcp

this is a template for creating personalized MCP servers. You can see larkin-mcp as an example, or check out my blog post about it here. It's technically a slightly more slimmed down version given I only really am trying to support uvx + pypi installs...

demo

https://github.com/user-attachments/assets/e96afd9e-786a-4394-91f9-9030f6ad2461 (sorry in advance about the low rez)

tech

this is using copier which is basically a more up to date cookiecutter + jinja project. I honestly hadn't heard of it before asking ChatGPT if there were alternatives or more up to date projects. But here we are.

what this creates

a MCP server that exposes your professional information (or really whatever you want to throw in there) to LLMs that have MCP connection (codex, claude code, claude desktop).

getting started

this should really be all you need to do:

without cloning:

$ YOURNAME='TODO' # PUT YOUR ACTUAL NAME HERE
$ brew install copier # (macos or linux... sorry windows)
$ copier copy gh:johnlarkin1/yourname-mcp $YOURNAME-mcp

with cloning:

$ brew install copier # or again uv tool install copier
$ YOURNAME='TODO' # PUT YOUR ACTUAL NAME HERE
$ git clone https://github.com/johnlarkin1/yourname-mcp.git
$ copier copy yourname-mcp my-mcp-project

answer questions!

Summary of what you're going to be asked for:

Variable	Description	Example
project_name	PyPI package name	johnsmith-mcp
person_full_name	Your full name	John Smith
person_first_name	First name (for prompts)	John
uri_scheme	Resource URI scheme	smith (creates smith://resume)
github_username	Your GitHub username	johnsmith
email	Your email	john@example.com
website_url	Your website (optional)	https://johnsmith.dev
location	Your location (optional)	New York, NY

populate markdown files

populate this stuff!

src/resources/content/
├── bio.md           # Your biography
├── contact.md       # Contact information (pre-filled from prompts)
├── projects.md      # Your projects
├── skills.md        # Technical skills
├── work.md          # Work experience
└── resume/
    └── resume.md    # Full resume

testing

cd my-mcp-project/$YOURNAME-mcp
uv sync
uv run pytest
uv run $YOURNAME-mcp

publishing

note, this is a bit more involved. Feel free to email me about it, or ask ChatGPT / Claude, but you'll want to set up a PyPi account if you want to actually publish it.

uv build
uv publish  # Requires PyPI token

using your local version (no pypi needed)

if you don't want to deal with pypi, you can just point claude desktop to your local project. the key is using uv run --directory to tell uv where your project lives:

{
  "mcpServers": {
    "yourname-mcp": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/absolute/path/to/your/yourname-mcp",
        "yourname-mcp"
      ]
    }
  }
}

so if your project is at ~/code/johnsmith-mcp, it'd look like:

{
  "mcpServers": {
    "johnsmith-mcp": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/Users/johnsmith/code/johnsmith-mcp",
        "johnsmith-mcp"
      ]
    }
  }
}

just make sure you've run uv sync in your project directory first!

using your published version

after you've published it, you can set this up in claude code or you can basically extend your claude_desktop.config. the config is cleaner since uvx fetches it automatically.

here's an example for mine:

╭─johnlarkin@Mac ~/Documents/coding/yourname-mcp-proj
╰─➤  vim ~/Library/Application\ Support/Claude/claude_desktop_config.json

you'll want a similar path. And then you'll have something like:

{
  "mcpServers": {
    "yourname-mcp": {
      "command": "uvx",
      "args": ["yourname-mcp"]
    }
  }
}

or for local (from within your project directory):

{
  "mcpServers": {
    "yourname-mcp": {
      "command": "uv",
      "args": ["run", "yourname-mcp"]
    }
  }
}

what's included:

Note

This section is Ai generated:

Resources

Your MCP server exposes these resources (via custom URI scheme):

• {scheme}://resume - Full resume markdown
• {scheme}://bio - Biography
• {scheme}://projects - Project portfolio
• {scheme}://contact - Contact information
• {scheme}://skills - Technical skills
• {scheme}://work - Work history

Tools

• get_resume(), get_bio(), get_projects(), etc. - Retrieve specific content
• search_info(query) - Search across all resources
• health_check() - Check server status

Prompts

Pre-built prompts for common use cases:

• summarize_for_role(role) - Tailored summary for a job role
• compare_to_job(job_description) - Analyze fit for a job
• interview_prep(role, company) - Interview preparation
• project_deep_dive(project_name) - Detailed project info