Lighthouse MCP Server: Setup, Features & Best Alternatives (2026)
Does Lighthouse Have an MCP Server?
As of February 2026, Lighthouse does not offer an official, first-party MCP server. This is the current landscape:
- Official MCP server: Not available from Lighthouse directly
- Community wrappers: Check GitHub for community-maintained projects that wrap the Lighthouse API for MCP compatibility
- API availability: Lighthouse offers a REST API that could be wrapped in a custom MCP server
- Expected timeline: No public announcements from Lighthouse about MCP support
The absence of an official server doesn’t mean you can’t use Lighthouse data with AI assistants. Several approaches exist, ranging from community wrappers to building your own MCP server around the Lighthouse API.
Why Connect Lighthouse to an MCP Server?
Lighthouse falls into the Testing & QA category of tools, providing test execution, browser automation, cross-browser testing, and performance analysis. Test management and debugging eat up development time. An MCP server for testing tools lets AI assistants run tests, analyze failures, and suggest fixes without switching between your editor, terminal, and browser-based test dashboards.
Practical Use Cases
- Run test suites and get AI-analyzed results without leaving your editor
- Debug test failures with AI-assisted stack trace analysis
- Generate test cases from specifications using natural language
- Query test coverage reports and identify untested code paths
- Monitor cross-browser test results and compatibility issues
Without an MCP server, each of these use cases requires manual context switching: open Lighthouse in a browser, find the data you need, copy it, paste it into your AI assistant, and wait for analysis. An MCP server automates that entire data-fetching step.
How to Set Up Lighthouse MCP (Community Approach)
Since Lighthouse doesn’t have an official MCP server, here’s how to set up a community wrapper or build your own.
lighthouse-mcp-server or mcp-lighthouse. Look for repositories with recent commits, good documentation, and active maintenance. Check the MCP server registry for officially listed community servers.If a community package exists, install it:
npm install -g @community/lighthouse-mcp-server
If building your own, clone a starter template:
git clone https://github.com/modelcontextprotocol/server-template
cd server-template
npm install
Add the server to your Claude Code MCP configuration at ~/.claude.json or your project’s .mcp.json:
{
"mcpServers": {
"lighthouse": {
"command": "npx",
"args": [
"@community/lighthouse-mcp-server"
],
"env": {
"LIGHTHOUSE_API_KEY": "your-api-key-here"
}
}
}
}
In Cursor, open Settings > MCP Servers and add the same configuration. Cursor uses the same MCP protocol as Claude Code, so most servers work identically in both environments.
{
"mcpServers": {
"lighthouse": {
"command": "npx",
"args": ["@community/lighthouse-mcp-server"],
"env": {
"LIGHTHOUSE_API_KEY": "your-api-key-here"
}
}
}
}
Restart your editor and test the connection by asking your AI assistant to query Lighthouse:
"Can you check if the Lighthouse MCP server is connected? Try listing available tools."
If the connection fails, check that your API key is valid, the MCP server process is running, and your firewall isn’t blocking local connections.
Lighthouse MCP Server: What You Can and Cannot Do
What Works with a Community Wrapper
Community-built MCP servers for Lighthouse typically support read operations – pulling data from the Lighthouse API and making it available to your AI assistant. This covers most analytical and reporting use cases.
| Operation | Typically Supported | Notes |
|---|---|---|
| Read data | Yes | Most wrappers expose read endpoints |
| Search/query | Yes | If the Lighthouse API supports it |
| Create/write | Sometimes | Depends on wrapper completeness |
| Delete/modify | Rarely | Most wrappers are read-only for safety |
| Webhooks/streaming | Rarely | Requires more complex server architecture |
| Authentication | Usually API key | OAuth flows are harder to implement in MCP |
Known Limitations
- Rate limits: Community wrappers inherit the rate limits of your Lighthouse API plan. Aggressive querying through an AI assistant can burn through API quota quickly.
- Data freshness: MCP servers query on demand, so data is as fresh as the API allows. There is no built-in caching in most community wrappers.
- Maintenance risk: Community projects may lag behind Lighthouse API changes. Check the repository’s last commit date before relying on it.
- Security: Storing API keys in MCP config files means they sit on disk in plain text. Use environment variables or a secrets manager for production setups.
Lighthouse MCP Pricing Considerations
The cost of using Lighthouse through MCP depends on several factors:
| Cost Component | Details |
|---|---|
| Lighthouse subscription | Your existing Lighthouse plan determines API access |
| API usage | Some plans have per-request or rate-limited API access |
| MCP server hosting | Free if running locally; cost if self-hosting remotely |
| Community wrapper | Usually free and open-source |
Important: Some Lighthouse plans restrict API access to higher tiers. Verify that your current plan includes the API endpoints the MCP wrapper needs before investing time in setup.
Best MCP Alternatives to Lighthouse
If Lighthouse doesn’t have a viable MCP server, or if the setup cost isn’t worth it for your use case, these alternatives provide similar capabilities through production-ready MCP servers.
Delulu9 – Best for Keyword Research MCP
If your primary need is keyword research, Delulu9 is purpose-built for MCP from day one. It isn’t a wrapper around another API – it is a native MCP server with its own data infrastructure.
| Feature | Delulu9 | Lighthouse (via wrapper) |
|---|---|---|
| Official MCP server | Yes (hosted, maintained) | Community-maintained |
| Setup time | Under 2 minutes | 15-60 minutes |
| Google keyword data | Yes | Depends on wrapper |
| Bing keyword data | Yes | Unlikely |
| Reddit keyword data | Yes | No |
| Search intent categorization | Built-in (automatic) | Manual or unavailable |
| Keyword difficulty scoring | Built-in | Depends on Lighthouse API |
| Rate limiting | 1,000 requests/hour | Depends on Lighthouse plan |
| Uptime & reliability | 99.9% (hosted infrastructure) | Depends on your setup |
| Price | $12/mo flat | Lighthouse subscription + API costs |
| Support | Email + documentation | Community only |
Why Delulu9 Is Different
Most MCP keyword research servers are thin wrappers around third-party APIs. Delulu9 is a vertically integrated tool: it collects data from Google, Bing, and Reddit, processes it through its own intent classification and difficulty scoring pipeline, and serves it through a hosted MCP server.
This matters because:
- No API key chain: You don’t need a separate subscription to another service. One API key, one bill.
- Reddit data is exclusive: No other keyword research MCP server combines Google, Bing, and Reddit autocomplete data in a single query.
- Intent is automatic: Every keyword result includes intent classification (informational, commercial, transactional, navigational) without manual tagging.
- It actually works: The server is hosted, monitored, and maintained. You don’t debug connection issues at 2 AM.
Quick Setup: Delulu9 MCP
{
"mcpServers": {
"delulu9": {
"command": "npx",
"args": ["-y", "@delulu9/seo-mcp-server@latest"],
"env": {
"DELULU9_API_KEY": "your-api-key"
}
}
}
}
That’s it. Restart your editor and ask Claude: “Find keyword ideas for equestrian using Delulu9.”
How to Choose Between Lighthouse MCP and Alternatives
- If you already pay for Lighthouse and need its specific data in your AI workflow, invest the time to set up a community MCP wrapper
- If keyword research is your primary need, Delulu9 at $12/mo is faster to set up and more reliable than wrapping another tool’s API
- If you need Lighthouse for non-keyword tasks (test execution, browser automation, cross-browser testing, and performance analysis), check whether a community MCP wrapper exists and is actively maintained
- Always verify API access is included in your Lighthouse plan before attempting MCP setup
- Consider running community wrappers locally first to test reliability before depending on them
Building Your Own Lighthouse MCP Server
If no community wrapper meets your needs, you can build a custom MCP server for Lighthouse. Here’s the architecture:
Your Editor (Claude Code / Cursor / Windsurf)
|
v
MCP Protocol (stdio or SSE)
|
v
Your Custom MCP Server (Node.js or Python)
|
v
Lighthouse REST API
The MCP SDK is available in TypeScript and Python:
# TypeScript
npm install @modelcontextprotocol/sdk
# Python
pip install mcp
Key decisions when building:
- Which Lighthouse API endpoints to expose – start with the 2-3 you use most
- Caching strategy – avoid hitting Lighthouse’s API on every query
- Error handling – gracefully handle rate limits and API outages
- Authentication – use environment variables, never hardcode keys
Building a custom server takes 2-8 hours depending on Lighthouse’s API complexity. For keyword research, this rarely makes sense when Delulu9 exists at $12/mo.
Lighthouse MCP vs REST API: When Do You Need MCP?
Not every Lighthouse integration needs MCP. Here’s when each approach makes sense:
| Scenario | Use MCP | Use REST API directly |
|---|---|---|
| AI assistant needs Lighthouse data | Yes | Possible but awkward |
| Automated scripts/pipelines | MCP or REST | REST is simpler |
| Browser-based dashboards | No | Yes |
| One-off data pulls | Either works | REST is fine |
| Conversational data exploration | Yes | No |
| Multi-tool AI workflows | Yes | Too complex with REST |
MCP’s value is in the conversational interface. If you’re building traditional automation, the Lighthouse REST API works fine. If you want AI assistants to intelligently query and combine data from multiple sources, MCP is the right abstraction.
~/.claude.json. See the step-by-step setup guide above for detailed instructions.
Related Guides
Get Started with MCP Keyword Research
If keyword research is part of your workflow, skip the wrapper complexity. Delulu9 gives you a production-ready MCP server with Google, Bing, and Reddit data, intent classification, and difficulty scoring. $12/mo, 2-minute setup, works with Claude Code, Cursor, and Windsurf.
Start your free trial – 1 day free, no credit card required.
Try Delulu9 free for 1 day
Keyword research from Claude Code. Google + Bing + Reddit data. $12/mo after trial.
Start Free Trial