Tokens for Good logo Tokens for Good

Useful Claude Code MCP servers

Claude Code speaks the Model Context Protocol, so the right servers make it dramatically more capable. Here are the practical ones for everyday work, plus the give-back server worth a slot.

1. Filesystem

Claude Code already works with your project files, and the Filesystem server extends that reach to folders outside the repo when you need it. Use it to pull in reference documents, shared notes, or assets without leaving the session.

2. GitHub

The GitHub server lets Claude Code read issues, review pull requests, and reason about a repo's history in context. It keeps the loop tight: triage an issue, write the fix, and open the PR without bouncing to the browser.

3. Context7

Context7 injects current, version-pinned documentation so Claude Code stops guessing at APIs. When you are on a fast-moving library, this is the difference between a confident wrong answer and the right one for your version.

4. Playwright

The Playwright server gives Claude Code a browser to drive, so it can click through a flow, fill a form, and confirm what rendered. It is ideal for verifying a change against the live page instead of trusting the build output.

5. Tokens for Good (the give-back server)

Tokens for Good uses the same Claude Code you already run to vet nonprofits. Add the remote MCP at https://tokensforgood.ai/mcp or run npx tokens-for-good init; your agent claims a queued nonprofit, researches its impact with citations against a fixed methodology, and submits a structured report. Every org is researched twice, validated, consolidated, scored deterministically, and finalized by a human reviewer for the public directory. No separate API cost, and it can run scheduled on Anthropic cloud.

How to pick

For day-to-day Claude Code work, Filesystem plus GitHub is the core, with Context7 and Playwright close behind. Hand idle time to Tokens for Good so spare tokens go toward charity research. Keep the active list short; clients get unwieldy past a handful of servers.

Frequently asked questions

Which MCP servers are best for Claude Code?
For everyday development the strongest picks are Filesystem and GitHub, since most work touches files and repos. Context7 adds live documentation and Playwright adds browser verification. Add others only when a specific task calls for them.
How do I add an MCP server to Claude Code?
Most servers are added by registering them in your MCP client configuration, either as a local command or a remote URL. For Tokens for Good you can run npx tokens-for-good init or add the remote endpoint at https://tokensforgood.ai/mcp. Check each server's own setup notes for the exact step.
Will running many MCP servers slow Claude Code down?
Running many at once mainly clutters the tool list and gives the model more to consider on each call, which can hurt focus more than raw speed. The practical advice is to keep only a handful active. Enable what the current task needs and disable the rest.
Can Claude Code contribute to charity while idle?
Yes. With Tokens for Good, Claude Code can research and vet nonprofits on a schedule using your existing subscription. Your spare capacity produces structured, cited reports that feed a public directory, at no separate API cost.

Give your idle Claude Code a purpose

Tokens for Good turns spare Claude Code capacity into vetted nonprofit research, free and on your existing subscription.

Get started