agentleFS
naman10parikhnaman10parikh/wikimem/.claude/skills/mcp-setup/SKILL.md

mcp-setup

Skill8 starsChanged 28 days agoMITView on GitHub

Self-install MCP servers to expand Claude's capabilities. Read the catalog, install the server, wire it into .mcp.json, and test.

  • Reads credentials
  • Installs packages

17 other projects use a version of this, 3 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 6 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 13 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check these sources in order:

    line 16
  2. Very specific:

    Read the current .mcp.json:

    line 97
  3. Distinctive:

    Add the new server entry. Preserve all existing entries.

    line 103
  4. Distinctive:

    Test by invoking one of the server's tools

    line 108
  5. Very specific:

    If it fails: check logs, verify env vars, try npx directly

    line 109
  6. Very specific:

    Add the new server to resources/awesome-repos.md under a "Currently Installed" section.

    line 113
  7. Very specific:

    NEVER install servers that require sudo or root access

    line 129
  8. Distinctive:

    NEVER pass secrets directly in args — use env vars

    line 130
  9. Distinctive:

    NEVER install from untrusted sources (verify GitHub stars, maintainer)

    line 131
  10. Distinctive:

    Always test in isolation before wiring into the main config

    line 132

“Common” is measured: kept by at least 6 in 10 of the 13 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

When to Use

  • User says "install the [X] MCP server" or "set up [X] MCP"
  • You need a capability that an MCP server provides (database access, API, browser, etc.)
  • During resource integration when an article mentions a useful MCP server

Process

1. Find the Server

Check these sources in order:

2. Evaluate Before Installing

  • Does it duplicate an existing capability? (Check current .mcp.json first)
  • Is it actively maintained? (Last commit < 3 months ago)
  • Does it have security concerns? (Never install servers that require write access to system files)
  • Is it worth the token overhead? (~4,200 tokens per MCP server vs ~70 for a skill)

Rule: If a skill can do the job, prefer the skill. MCPs are for external integrations only.

3. Install

For npx-based servers (most common):

# Test it works first
npx -y @modelcontextprotocol/server-[name] --help

Then update .mcp.json:

{
  "mcpServers": {
    "[name]": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-[name]"],
      "env": {}
    }
  }
}

For pip-based servers:

pip install mcp-server-[name]

Then update .mcp.json:

{
  "mcpServers": {
    "[name]": {
      "command": "python",
      "args": ["-m", "mcp_server_[name]"],
      "env": {}
    }
  }
}

For Docker-based servers:

docker pull mcp/[name]

Then update .mcp.json:

{
  "mcpServers": {
    "[name]": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "mcp/[name]"],
      "env": {}
    }
  }
}

4. Wire Into Settings

Read the current .mcp.json:

cat $PROJECT_ROOT/.mcp.json

Add the new server entry. Preserve all existing entries.

5. Verify

  • Restart Claude Code session (or the server auto-discovers on next tool call)
  • Test by invoking one of the server's tools
  • If it fails: check logs, verify env vars, try npx directly

6. Document

Add the new server to resources/awesome-repos.md under a "Currently Installed" section.

Common MCP Servers Worth Having

ServerWhat It DoesInstall
filesystemRead/write files with permissionsnpx -y @modelcontextprotocol/server-filesystem
githubGitHub API (PRs, issues, repos)npx -y @modelcontextprotocol/server-github
postgresQuery PostgreSQL databasesnpx -y @modelcontextprotocol/server-postgres
fetchHTTP requests with smart renderingnpx -y @modelcontextprotocol/server-fetch
memoryKey-value persistent memorynpx -y @modelcontextprotocol/server-memory
puppeteerBrowser automationnpx -y @modelcontextprotocol/server-puppeteer
firecrawlWeb scraping and crawlingnpx -y firecrawl-mcp (requires FIRECRAWL_API_KEY)

Security Rules

  • NEVER install servers that require sudo or root access
  • NEVER pass secrets directly in args — use env vars
  • NEVER install from untrusted sources (verify GitHub stars, maintainer)
  • Always test in isolation before wiring into the main config

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.