> ## Documentation Index
> Fetch the complete documentation index at: https://docs.0mcp.io/llms.txt
> Use this file to discover all available pages before exploring further.

# 10 useful MCP servers for developers

> Compare ten maintained MCP servers for source control, browser automation, payments, docs, observability, databases, cloud, and local development.

The best [MCP server](https://modelcontextprotocol.io/) is the one that supports your task with the narrowest permissions, a trusted maintainer, clear documentation, and an active release process.

This list highlights ten useful starting points. It is not a universal ranking. Server features and availability change, so verify the linked primary documentation before production use.

## Quick comparison

| MCP server      | Best for                                | Delivery                 |
| --------------- | --------------------------------------- | ------------------------ |
| GitHub          | Repositories, issues, and pull requests | Local and remote options |
| Playwright      | Browser automation and web testing      | Local                    |
| Stripe          | Payments and Stripe documentation       | Remote and local options |
| Notion          | Workspace search and page workflows     | Hosted remote            |
| Sentry          | Issue and application diagnostics       | Remote and local options |
| Cloudflare      | Cloudflare configuration and operations | Managed remote           |
| Atlassian Rovo  | Jira and Confluence workflows           | Hosted remote            |
| Google BigQuery | Governed analytics and SQL tasks        | Managed remote           |
| Filesystem      | Controlled local file operations        | Local reference server   |
| Fetch           | Web content retrieval                   | Local reference server   |

## 1. GitHub MCP Server

[GitHub's official MCP server](https://github.com/github/github-mcp-server) connects AI applications to repositories, issues, pull requests, code, and other GitHub workflows.

Use it when an agent needs structured GitHub context or approved repository actions. Restrict available toolsets and token permissions to the required repositories and operations.

## 2. Playwright MCP

[Microsoft Playwright MCP](https://github.com/microsoft/playwright-mcp) provides browser automation through structured page information and browser actions.

It is useful for exploratory automation, web testing, and workflows that need persistent browser state. Treat browser access as sensitive because pages can contain untrusted content and authenticated sessions.

## 3. Stripe MCP Server

[Stripe's MCP server](https://docs.stripe.com/mcp) provides tools for selected Stripe API operations and documentation search.

Use test mode during development. Apply narrow credentials, human confirmation, and audit logging before allowing financial or customer-data changes.

## 4. Notion MCP

[Notion MCP](https://developers.notion.com/guides/mcp/overview) is a hosted server that gives supported AI clients access to authorized Notion workspace content and actions.

It fits research, documentation, and workspace workflows. Review the pages and data sources available to the connected identity.

## 5. Sentry MCP

[Sentry MCP](https://github.com/getsentry/sentry-mcp) helps developers investigate issues, events, projects, and application diagnostics.

It is useful in incident triage and debugging. Limit access to the required organizations and projects, and treat stack traces and event data as potentially sensitive.

## 6. Cloudflare MCP servers

[Cloudflare's managed MCP servers](https://developers.cloudflare.com/agents/model-context-protocol/cloudflare/servers-for-cloudflare/) expose Cloudflare account and product workflows through remote MCP endpoints.

They can support configuration review and operational automation. Separate read-only investigation from actions that change production infrastructure.

## 7. Atlassian Rovo MCP Server

[Atlassian Rovo MCP](https://support.atlassian.com/atlassian-rovo-mcp-server/docs/getting-started-with-the-atlassian-remote-mcp-server/) connects supported clients to Jira, Confluence, Jira Service Management, Bitbucket, and related Atlassian context.

It suits project tracking, documentation search, and work-item updates. Access follows the connected user's permissions, but you should still confirm sensitive write operations.

## 8. Google BigQuery MCP server

[Google's BigQuery remote MCP server](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp) supports metadata discovery, resource listing, and governed query workflows.

Use it for analytical agents that need current warehouse data without copying entire datasets into model context. Control project, dataset, table, and job permissions through Google Cloud IAM.

## 9. Filesystem reference server

The [Filesystem server](https://github.com/modelcontextprotocol/servers/tree/main/src/filesystem) demonstrates secure local file operations within configured directories.

It is useful for development and learning. Allow only the smallest required roots, inspect the package source, and remember that a local server runs with the user's operating-system permissions.

## 10. Fetch reference server

The [Fetch server](https://github.com/modelcontextprotocol/servers/tree/main/src/fetch) retrieves web content in a format suitable for language models.

It is useful for learning and controlled retrieval workflows. Apply URL allowlists, redirect checks, private-network blocking, timeouts, and response-size limits to reduce SSRF and data-exfiltration risks.

<Warning>
  The MCP project's reference servers are educational examples, not a promise of production readiness. Evaluate each server against your own threat model.
</Warning>

## How to evaluate any MCP server

Before installation, check:

1. **Publisher:** Is the package or endpoint controlled by the expected organization?
2. **Source:** Can you review the code, license, releases, and security policy?
3. **Permissions:** Can you limit [tools](/learn/core-concepts/tools), accounts, scopes, hosts, and file paths?
4. **Transport:** Is the local command or remote endpoint documented and secure?
5. **Maintenance:** Are releases, dependencies, and vulnerability reports handled?
6. **Data handling:** What reaches the server, model provider, logs, and upstream service?
7. **Change control:** Can you pin, test, approve, and roll back updates?

Use the [official MCP Registry](https://registry.modelcontextprotocol.io/) for discovery, then confirm ownership on the publisher's official website or repository.

## When to build your own server

Build or generate a dedicated server when:

* No trusted server supports your API.
* You need task-specific tools rather than a broad integration.
* Data residency or self-hosting is required.
* You need custom authorization, approval, or audit rules.
* You want to expose only selected API operations.

[0mcp](https://0mcp.io) can create a hosted MCP server from OpenAPI 3.x, Swagger 2.0, a direct REST API configuration, or GraphQL. Each source type supports tools, resources, prompts, Playground testing, versioning and rollback, analytics, logs, and client setup guidance.

## Key takeaway

**Choose MCP servers by task fit, publisher trust, permissions, security controls, and maintenance—not by the size of a public list.**