Recommended Free Tools
Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Use claude mcp add --transport http <name> <url> for a hosted MCP server, or claude mcp add <name> -- <command> [args...] for a local stdio server. Then choose the right configuration scope, complete the server’s authentication flow, and verify the connection with claude mcp list, claude mcp get <name>, or Claude Code’s /mcp panel. MCP (Model Context Protocol) is the open standard that lets Claude Code use tools and data supplied by external servers.
What MCP means in Claude Code
In this setup, Claude Code is the MCP client. An MCP server publishes tools, databases, resources, or prompts; Claude Code discovers those capabilities and can request them during a session. The Model Context Protocol documentation describes MCP as “an open-source standard for connecting AI applications to external systems.”
Capabilities depend on the server. An issue-tracker server might read and update tickets, a monitoring server might query incidents, and a database server might run approved queries. Do not assume a server can perform an operation until its own documentation lists the relevant tool and permissions.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Before you add a server
- Install and launch a current Claude Code CLI, then run the command from the project or shell context where you intend to use the server.
- Obtain the provider’s official MCP endpoint or local launch command. Instructions written for another MCP client usually contain an
mcpServersJSON entry that can be translated. - Identify the documented transport: HTTP, stdio, SSE, or WebSocket. Prefer HTTP for a hosted server when it is available.
- Decide whether the server should be private to one project, shared through a project file, or available across your projects.
- Review the operator, requested tools, credentials, and data access. Anthropic’s guidance is direct: “Verify you trust each server before connecting it.” A server that fetches external content can expose you to prompt-injection risk.
Connect a remote MCP server over HTTP
For a hosted endpoint, use the HTTP transport explicitly:
#1 Best Overall
claude mcp add --transport http notion https://mcp.notion.com/mcp
Replace notion with a short name you will recognize and replace the URL with the server’s exact endpoint. HTTP is the current reference’s recommended choice for remote MCP services.
Authenticate after adding
If the service supports OAuth, open Claude Code’s /mcp panel and complete its sign-in flow. Other services may require headers, an API key, or provider-specific credentials. Use the provider’s current names and scopes; never paste a live secret into a shell history, a committed file, or an article example.
Check that configuration and health are different
The CLI’s “Added” response means configuration was written. It does not prove that the endpoint is reachable or that your account is authorized. Run:
claude mcp list
claude mcp get notion
Then open /mcp inside Claude Code to inspect status, authenticate, and review available tools.
Run a local MCP server with stdio
Use stdio when Claude Code should start a local process, such as a package installed with Node.js:
claude mcp add --transport stdio example -- npx -y @example/mcp-server
The -- separator is essential. Everything after it belongs to the server command and its arguments; everything before it is interpreted by Claude Code.
Pass an environment variable without exposing it as an argument
claude mcp add --env API_KEY=your-key --transport stdio example -- npx -y @example/mcp-server
Use a placeholder while testing and provide the real value through your operating system’s secret-management method. Confirm that the runtime and executable are installed and available on PATH.
Free tools Windows power users keep installed
One-click scans. No signup required.
Rank #2
Native Windows considerations
Shell quoting and executable resolution differ on Windows. Follow the current Claude Code documentation for your shell when using commands such as npx; if a command works in one terminal but not another, first check the resolved executable path and environment variables.
Choose the right configuration scope
Scope controls who can see a server and where its definition is stored.
| Scope | Use it when | Storage and review |
|---|---|---|
| Local | You need a private definition for the current project and user context. | Claude Code stores local-scoped configuration per project in ~/.claude.json. |
| Project | A team should share the server definition through version control. | Stored in the project-root .mcp.json. Keep secrets out; use environment variables or a credential manager. Interactive sessions request approval before using project-scoped servers. |
| User | You want the server available across your projects but private to your account. | Managed as a user-level definition rather than a shared project file. |
When a server exists in multiple scopes, the documented precedence is local, then project, then user. Claude Code uses the complete higher-priority definition; it does not merge individual fields from lower-priority entries. Plugin servers and Claude.ai connectors participate in the wider hierarchy. These behaviors can change with CLI versions, so confirm them in the current MCP reference.
Project JSON example
A project file uses an mcpServers object. A remote entry needs a type such as http:
{
"mcpServers": {
"team-service": {
"type": "http",
"url": "https://service.example/mcp"
}
}
}
The hostname above is illustrative; use the real URL supplied by your server operator. A remote URL without a valid type is a configuration error in the current reference.
Translate another client’s MCP configuration
If a provider gives you JSON for another MCP client, copy the contents of its mcpServers object and either add it with:
claude mcp add-json team-service '{"type":"http","url":"https://service.example/mcp"}'
or adapt the entry in a project .mcp.json. Check the transport before translating: remote services use http, sse, or ws as documented; local programs use stdio-style command and args.
Rank #3
Less-common transports: SSE and WebSocket
SSE
Server-Sent Events is deprecated in the current reference. Some providers still expose only SSE, in which case use the explicitly documented form for your installed Claude Code version:
claude mcp add --transport sse service https://service.example/sse
Choose HTTP instead whenever the provider offers an HTTP MCP endpoint.
WebSocket
The --transport flag does not accept ws in the current reference. Configure a WebSocket server with claude mcp add-json or a .mcp.json entry whose type and URL follow the provider’s instructions. WebSocket is appropriate for a service that needs a persistent, bidirectional connection; an ordinary request/response service is usually simpler over HTTP.
Authenticate with the minimum access
Remote authentication can use OAuth through /mcp, authorization headers, or server-specific API credentials. Follow the server’s current setup page for exact header names, client IDs, callback ports, client secrets, and scopes. Store credentials outside committed JSON and shell history. For a shared project file, commit only the endpoint and non-sensitive defaults; inject secrets at runtime.
After login, perform a small read-only request where possible. Confirm the expected tool appears and that its returned data is appropriate before allowing writes, deletes, code execution, or access to sensitive systems.
Verify and use the server safely
- Run
claude mcp listand confirm the server’s status. - Run
claude mcp get <name>to inspect its URL, transport, scope, and configuration details. - Open
/mcpin Claude Code to complete authentication and review tools. - Ask Claude for a low-risk, read-only operation and check the tool name and returned fields.
- Approve project-scoped servers only after reviewing the checked-in
.mcp.json.
Treat content returned from tools as untrusted input when it originates on the open web. A page, ticket, or document can contain instructions aimed at changing Claude’s behavior rather than useful task data.
Or skip the browser setup: ScreenshotNeo through MCP
ScreenshotNeo is a website screenshot API and MCP server for Claude, Cursor, and other MCP clients. Its MCP tools—take_screenshot, get_page_info, and capture_pdf—let an agent capture pages without you wiring up a browser automation process. Connect it using the provider’s current MCP instructions, then verify it in /mcp like any other remote server.
Rank #4
If you prefer a direct HTTP call, the API returns PNG, JPEG, WebP, or PDF:
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webp
Python:
import requests
r = requests.get("https://api.screenshotneo.com/v1/shot", params={"access_key": "YOUR_API_KEY", "url": "https://stripe.com"}, timeout=90)
open("shot.webp", "wb").write(r.content)
Node.js:
const q = new URLSearchParams({ access_key: 'YOUR_API_KEY', url: 'https://stripe.com' });
const res = await fetch(`https://api.screenshotneo.com/v1/shot?${q}`);
See the ScreenshotNeo documentation for authentication and options. Before capture, it can accept cookie or consent banners and remove more than 60 known consent platforms, newsletter popups, and chat widgets; each cleanup step can be disabled. Bot checks or CAPTCHAs, blank pages, timeouts, failed loads, and cache hits are not billed, and response headers report the page verdict and billing result.
It also supports full-page lazy-image capture, CSS-selector elements, dark mode, device presets, retina scale, PDF paper settings and page ranges, custom CSS and JavaScript, clicks, waits, request blocking, headers, cookies, user agents, authorization, timezone, geolocation, transparent backgrounds, resizing, chosen cache TTLs, signed image links, asynchronous jobs with signed webhooks, bulk capture of 100 URLs per call, usage reporting, and an OpenAPI specification. Its parameter names are compatible with those used by other screenshot APIs, which can simplify migration.
Every plan includes every feature: 1,000 screenshots per month are free with no card; paid plans start at $5 for 3,000 shots. Create a free ScreenshotNeo account.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Troubleshooting MCP connections
“Added” appears, but the server is unavailable
Adding writes configuration only. Run claude mcp list for health and claude mcp get <name> for details. Check DNS, firewall rules, endpoint spelling, and provider status.
A local process exits immediately
Verify the runtime or package is installed, the executable is on PATH, and every server argument follows --. Run the underlying command independently to reveal package or permission errors.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11The remote service asks you to log in
Open /mcp and complete the supported OAuth flow. Confirm that the account has access to the requested workspace and scopes.
A project server is waiting for approval
Open Claude Code in that project, inspect .mcp.json, and approve only the servers and capabilities you trust.
Best Value
JSON will not load
Validate JSON syntax, include a correct type, and provide either a valid remote URL or the required local command and args. A URL without a type is rejected by the current configuration rules.
The transport does not match
Confirm the provider’s endpoint type. Prefer HTTP over deprecated SSE; configure WebSocket through JSON rather than the --transport flag.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Operational and cost considerations
Remote HTTP avoids maintaining a local runtime but requires network availability and trust in the service operator. Stdio gives local system access and can work offline, but you must patch the runtime and package and protect its environment variables. Project scope improves team consistency while increasing review and secret-management responsibility. Start with read-only tools and narrow scopes, then expand only when the workflow requires it.
Claude Code documentation includes output warning and maximum-token settings (10,000 and 25,000 tokens in the referenced behavior); these are software defaults that may change with the CLI version, not service-level guarantees. Re-check the current reference before relying on them.
Frequently Asked Questions
Can I add more than one MCP server to Claude Code?
Yes. Give each server a distinct name, add it with its documented transport, and inspect the combined status with claude mcp list or /mcp.
Should I use HTTP or stdio?
Use HTTP for a hosted endpoint and stdio when Claude Code should launch a local command. The server provider’s supported transport is the deciding factor.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsWhere do I see the tools a server exposes?
Open the /mcp panel after connecting. It is the place to authenticate and review the server’s available capabilities.
Is SSE still supported?
Some services may still require it, but the current Claude Code reference marks SSE as deprecated and recommends HTTP where available.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.

