Integrates Lighthouse to perform Core Web Vitals analysis, allowing users to analyze URL performance, compare metrics between pages, verify against performance thresholds, and crawl sites for comprehensive auditing.
Part of the TestScout MCP Suite — AI-powered QA tools for modern development teams.
Core Web Vitals analysis powered by Lighthouse. Runs as an MCP server for AI coding tools (Claude Code, Cursor, Windsurf, etc.) or as a standalone CLI for CI pipelines and terminals. Four tools: analyze a URL, compare two URLs, check against thresholds, or crawl an entire site.
Node.js >= 18
Google Chrome or Chromium
Related MCP server: Lighthouse MCP
claude mcp add lightscout -- npx -y lightscout-mcpAll use the same mcpServers JSON format. Add this to the appropriate config file:
{
"mcpServers": {
"lightscout": {
"command": "npx",
"args": ["-y", "lightscout-mcp"]
}
}
}Client | Config file |
Cursor |
|
VS Code (Copilot) |
|
Windsurf |
|
Cline | Settings > MCP > Edit Config |
Roo Code |
|
Gemini CLI |
|
Zed uses context_servers instead of mcpServers:
{
"context_servers": {
"lightscout": {
"command": "npx",
"args": ["-y", "lightscout-mcp"]
}
}
}npm install -g lightscout-mcpnpx -p lightscout-mcp lightscout analyze https://example.comlightscout analyze <url> [--device mobile|desktop] [--categories perf,a11y,seo,best-practices]
lightscout compare <urlA> [urlB] [--device mobile|desktop]
lightscout crawl <url> [--max-pages N] [--device mobile|desktop]
lightscout check <url> [--perf 90] [--lcp 2500] [--fcp 3000] [--cls 0.1] [--tbt 300] [--ttfb 1800]Analyze a single page:
lightscout analyze https://example.com
lightscout analyze https://example.com --device desktop --categories performance,accessibilityCompare two URLs, or the same URL on mobile vs desktop:
lightscout compare https://a.com https://b.com
lightscout compare https://example.com # mobile vs desktopCrawl a site and audit discovered pages:
lightscout crawl https://example.com --max-pages 5CI quality gate (exits 1 on failure):
lightscout check https://example.com --perf 90 --lcp 2500Run Lighthouse on a URL. Returns performance scores, Core Web Vitals (LCP, FCP, CLS, TBT, SI, TTFB) with good/needs-improvement/poor ratings, top recommendations sorted by impact, and diagnostics.
Parameter | Type | Required | Description |
| string | yes | URL to analyze |
|
| no | Device emulation (default: mobile) |
| string[] | no | Lighthouse categories (default: |
Compare Core Web Vitals between two URLs, or the same URL on mobile vs desktop. Shows per-metric deltas with winner indicators.
Parameter | Type | Required | Description |
| string | yes | First URL |
| string | no | Second URL. If omitted, compares urlA mobile vs desktop |
|
| no | Device emulation when comparing two URLs (default: mobile) |
Pass/fail check against performance thresholds. Useful for CI quality gates.
Parameter | Type | Required | Description |
| string | yes | URL to check |
|
| no | Device emulation (default: mobile) |
| object | no | Custom thresholds (see defaults below) |
Default thresholds use Google's "poor" boundary -- only fails if metrics are genuinely bad:
Metric | Default (fail if worse) |
performance | < 50 |
LCP | > 4000ms |
FCP | > 3000ms |
CLS | > 0.25 |
TBT | > 600ms |
TTFB | > 1800ms |
Discover pages on a site (via sitemap.xml or link crawling) and run Lighthouse on each. Returns per-page results plus site-wide summary with avg/min/max scores and common issues.
Parameter | Type | Required | Description |
| string | yes | Site URL or homepage to crawl |
|
| no | Device emulation (default: mobile) |
| number | no | Maximum pages to analyze (default: 20, max: 20) |
Pages are discovered via sitemap.xml first, falling back to link crawling. Up to 3 Lighthouse runs execute in parallel.
Metric | Good | Needs Improvement | Poor |
LCP | <= 2500ms | 2500-4000ms | > 4000ms |
FCP | <= 1800ms | 1800-3000ms | > 3000ms |
CLS | <= 0.1 | 0.1-0.25 | > 0.25 |
TBT | <= 200ms | 200-600ms | > 600ms |
SI | <= 3400ms | 3400-5800ms | > 5800ms |
TTFB | <= 800ms | 800-1800ms | > 1800ms |
TBT (Total Blocking Time) is used instead of INP/FID because Lighthouse runs lab-only tests with no real user interaction.
Up to 3 concurrent Lighthouse runs (semaphore-controlled).
Chrome is launched headless and killed after each audit.
MCP responses are token-optimized with condensed metrics and compact JSON.
LightScout is the first tool in the TestScout MCP Suite. More tools are coming:
Tool | Description | Status |
lightscout-mcp | Core Web Vitals & Lighthouse analysis | Available |
testscout-diagnose | Error triage & root cause analysis | Coming soon |
testscout-scrape | Structured data extraction for testing | Coming soon |
testscout-plan | AI test plan & code generation | Coming soon |
testscout-load | Load test generation & analysis | Coming soon |
testscout-maintain | Self-healing test maintenance | Coming soon |
Follow development at testscout.dev.
MIT
Run Lighthouse on a URL and get Core Web Vitals scores, metrics, and prioritized recommendations. Returns LCP, FCP, CLS, TBT, SI, TTFB with good/needs-improvement/poor ratings.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | URL to analyze | |
| device | No | Device emulation (default: mobile) | |
| categories | No | Lighthouse categories: performance, accessibility, seo, best-practices (default: [performance]) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden. It discloses the outputs (LCP, FCP, CLS, etc.) and that it runs Lighthouse, but does not mention side effects, auth requirements, rate limits, or destructive potential. It is sufficient for a read-only analysis but lacks detail on behavioral traits like execution time or failure modes.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences with no filler. The first sentence front-loads the key action ('Run Lighthouse on a URL') and output, the second lists specific metrics. Every sentence is purposeful and concise.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description covers the main outputs but lacks details on execution context (e.g., time to run, caching, error handling). With no output schema or annotations, there are gaps such as how results are returned and potential limitations.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% (all parameters described). The description adds no additional meaning beyond the schema – it does not clarify formats, defaults, or constraints. Baseline of 3 is appropriate as per guidelines.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states it runs Lighthouse on a URL and returns Core Web Vitals, metrics, and recommendations. The verb 'Run Lighthouse' and resource 'URL' are specific, and the output is detailed. It distinguishes itself from siblings like 'crawl_site' by focusing on a single page analysis.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives (check_threshold, compare_performance, crawl_site). There is no mention of context, prerequisites, or situations to avoid. The agent must infer usage from the purpose alone.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Pass/fail check against performance thresholds. Use for CI quality gates. Defaults use Google's 'poor' thresholds — only fails if metrics are genuinely bad.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | URL to check | |
| device | No | Device emulation (default: mobile) | |
| thresholds | No | Custom thresholds. Defaults to Google's 'poor' boundary. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the burden. It mentions defaults use Google's 'poor' thresholds and that it only fails on genuinely bad metrics, but does not explain the return format or side effects (though likely read-only).
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences effectively convey purpose, usage, and default behavior with no waste. Front-loaded for quick agent scanning.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given 100% schema coverage and no output schema, the description covers important aspects: purpose, usage context, and default behavior. Missing explicit return format but adequate for a simple pass/fail tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so baseline is 3. The description adds that defaults are Google's 'poor' boundaries, which is already in the thresholds parameter description, providing minimal extra value.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool performs a 'pass/fail check against performance thresholds' for CI quality gates, differentiating it from sibling tools like analyze_performance and compare_performance.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It explicitly recommends use for 'CI quality gates,' providing clear context. It does not mention when not to use, but the usage is well-defined.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Compare Core Web Vitals between two URLs, or the same URL on mobile vs desktop. Shows per-metric deltas with winner indicators.
| Name | Required | Description | Default |
|---|---|---|---|
| urlA | Yes | First URL to compare | |
| urlB | No | Second URL to compare. If omitted, compares urlA mobile vs desktop. | |
| device | No | Device emulation when comparing two different URLs (default: mobile) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations exist, so the description must carry the full burden. It effectively communicates that this is a read-only comparison tool showing deltas and winners. However, it lacks details on data freshness, caching, or prerequisites (e.g., CrUX data availability), which would have raised the score.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two well-structured sentences. The first sentence defines the action and scope, the second specifies the output format. Every word is necessary, and the description is front-loaded with critical information.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the moderate complexity and absence of output schema and annotations, the description covers the core functionality well but omits details like the specific metrics (LCP, FID, CLS), data source, and whether it works for any URL or only those with CrUX data. These gaps reduce completeness.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, and the description adds value by explaining the behavior when urlB is omitted (compares urlA mobile vs desktop) and the default device. This goes beyond the schema's parameter descriptions, providing practical usage instructions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool compares Core Web Vitals between two URLs or the same URL on different devices, with a specific output of per-metric deltas and winner indicators. It distinguishes itself from siblings like analyze_performance, which likely handles single URLs.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides clear use cases: comparing two URLs or the same URL across devices. It explains the fallback when urlB is omitted, giving implicit guidance on when to include it. While it doesn't explicitly name alternatives, the sibling tools provide context for when not to use this tool.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Discover all pages on a site (via sitemap.xml or link crawling) and run Lighthouse on each. Returns per-page results plus site-wide summary with avg/min/max scores and common issues. Max 20 pages, 3 parallel runs.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | Site URL or homepage to crawl | |
| device | No | Device emulation (default: mobile) | |
| maxPages | No | Maximum pages to analyze (default: 20, max: 20) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Discloses limits (max pages, parallel runs) and output structure, but with no annotations, it could mention more traits like potential slowness or authentication needs.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Three concise sentences: purpose, output, constraints. No wasted words, front-loaded.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Covers purpose, output, and constraints adequately for a tool with few parameters and no output schema. Could include error handling or behavior on missing sitemap.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema covers all 3 parameters (100% coverage), so description adds little beyond clarifying crawling method for url. Baseline 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states it discovers all pages via sitemap or crawling and runs Lighthouse, returning per-page and summary results. This distinguishes it from siblings that focus on single pages or comparisons.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Provides context (site-wide audit) and constraints (max 20 pages, 3 parallel runs), but does not explicitly state when to use vs alternatives like analyze_performance for single pages.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v1.0.1analyze_performancecheck_thresholdcompare_performancecrawl_siteScored across 4 tools
Each tool has a clearly distinct purpose: single URL analysis, threshold pass/fail check, two-URL comparison, and multi-page crawl. No overlap or ambiguity.
All tools follow a consistent verb_noun pattern with lowercase underscores: analyze_performance, check_threshold, compare_performance, crawl_site. No deviations.
Four tools is well-scoped for the performance auditing domain. Each tool serves a necessary function without bloat or deficiency.
Core workflows (single analysis, comparison, crawl, CI gate) are covered. Minor gap: no tool for customizing thresholds, but defaults are sensible.
Run SEO + AI-visibility (GEO) audits from Claude, Cursor & other AI clients.
Analyze web performance and get optimization insights from GTmetrix, directly in your AI workflow.
Live SEO workflow tools for Claude Code, Codex, and AI agents.
Screenshot, diff, audit and sitemap-capture any web page — 5 MCP tools for AI agents.
