PageLens AI
Your AI assistant, plugged into your PageLens audits.
PageLens AI — MCP Server
AI agents can run senior-level website audits in seconds.
An MCP (Model Context Protocol) server that gives AI agents direct access to PageLens AI — automated website reviews covering UX, SEO, Performance, Accessibility, Security, and Conversion.
Plug it into Cursor, Claude, or any MCP-compatible client and your agent can read scan results, drill into findings, surface quick wins, and close the fix loop — all without leaving the IDE.
MCP Endpoint
https://www.pagelensai.com/api/mcp
This is a remote MCP server — no local install required.
Quick Start
Cursor
Add to your ~/.cursor/mcp.json (or workspace .cursor/mcp.json):
{
"mcpServers": {
"pagelens": {
"url": "https://www.pagelensai.com/api/mcp"
}
}
}
Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": {
"pagelens": {
"url": "https://www.pagelensai.com/api/mcp"
}
}
}
Authentication
This server uses OAuth 2.0. Your MCP client will open a browser window to authorize with your PageLens account on first connect. No API keys to manage.
Scopes granted on authorization:
| Scope | What it unlocks |
|---|---|
read:scans | List and read your scan results |
read:findings | Read findings and quick wins |
write:feedback | Submit finding feedback and owner decisions |
Available Tools
whoami
Confirm which PageLens account this MCP session is operating on, plus granted OAuth scopes.
{}
list_domains
List the domains you've verified ownership of, along with badge tier and the scan currently anchored to the public badge.
{
"include_unverified": false
}
list_scans
List your most recent PageLens scans. Filter by status, domain, or date. Returns a slim summary per scan (id, URL, score, grade, severity counts).
{
"limit": 20,
"status": "COMPLETE",
"domain": "example.com",
"since": "2026-01-01T00:00:00Z"
}
Status values: PENDING · RUNNING · COMPLETE · FAILED · CANCELLED
get_scan
Read the full summary for a single scan: score, grade, severity counts, executive summary, top-5 highest-priority findings, and per-persona reviews.
{
"scan_id": "clxxxxxxxxxxxxxxx"
}
list_findings
Page through all findings for a scan. Filter by severity, category, persona, page URL, or rule ID. Use format: "full" to include descriptions, suggestions, and evidence.
{
"scan_id": "clxxxxxxxxxxxxxxx",
"severity": "HIGH",
"category": "SECURITY",
"format": "full",
"limit": 50
}
Severity levels: CRITICAL · HIGH · MEDIUM · LOW · INFO
Categories: UX · SEO · PERFORMANCE · ACCESSIBILITY · SECURITY · CONTENT · HEADERS · DESIGN · ERROR
Personas: MARKETER · CRO · UX · ACCESSIBILITY · BRAND · EXECUTIVE · PERFORMANCE · SEO
get_quick_wins
Return the top N quick-win findings — high impact, low-to-moderate effort — ranked by the same Impact × Effort scorer used in the PageLens dashboard. Optionally override the scan's preset to re-rank under a different lens.
{
"scan_id": "clxxxxxxxxxxxxxxx",
"limit": 5,
"preset_override": "CONVERSION"
}
Presets: PRE_SALES · PRE_LAUNCH · CONVERSION · INVESTOR · BRAND_POLISH
report_finding_feedback
Flag a finding as a false positive, wrong severity, wrong category, or not actionable. Requires a paragraph of reasoning and a concrete evidence snippet.
{
"finding_id": "clxxxxxxxxxxxxxxx",
"kind": "FALSE_POSITIVE",
"reason": "The selector .pointer-events-auto matches 100+ utility classes, not a single offending element.",
"evidence": "<div class=\"pointer-events-auto ...\"> — Tailwind utility, not an event-handler.",
"proposed_severity": "LOW"
}
Feedback kinds: FALSE_POSITIVE · INCORRECT_SEVERITY · INCORRECT_CATEGORY · NOT_ACTIONABLE · OTHER
acknowledge_finding_decision
Attach owner-controlled context to a finding when the issue is real, but reflects an intentional architecture, security, or product tradeoff.
This does not hide the finding, edit the report, or change the PageLens score. It records the rationale so the report can show that the owner has acknowledged the tradeoff, and future scans can recognise the same finding as previously acknowledged.
{
"finding_id": "clxxxxxxxxxxxxxxx",
"decision": "INTENTIONAL_TRADEOFF",
"reason": "Next.js Cache Components and PPR currently prevent us from using per-request CSP nonces safely.",
"evidence": "proxy.ts documents the CSP tradeoff: script-src uses 'unsafe-inline' because cached HTML shells cannot vary nonces per request.",
"expires_at": "2026-10-01T00:00:00Z"
}
Decision kinds: ACKNOWLEDGED · ACCEPTED_RISK · INTENTIONAL_TRADEOFF · WONT_FIX_NOW
clear_finding_decision
Clear a previously acknowledged decision so it stops appearing on current and future reports. The audit history is preserved.
{
"decision_id": "clxxxxxxxxxxxxxxx",
"reason": "We have migrated to a nonce-compatible rendering path."
}
You can also clear by finding_id when you do not have the decision_id.
What PageLens Checks
Each scan runs a deterministic rule engine + AI reviewer pipeline across every page:
| Area | Examples |
|---|---|
| SEO | Title/meta length, canonical URL, Open Graph, heading hierarchy |
| Performance | Core Web Vitals (LCP, CLS, INP), page weight, render-blocking resources, DOM size, third-party load |
| Security | HTTP→HTTPS redirect, exposed .env/.git files, SSL expiry, XSS surfaces, source maps, exposed secrets in page source |
| Accessibility | Focus-visible CSS, reduced-motion support, ARIA patterns |
| UX | Hero hierarchy, mobile menu patterns, CTA structure |
| Content | Placeholder text, stale copyright year, mixed-content references |
Findings include severity, effort estimate, copy-pasteable evidence, and a one-line fix suggestion.
Example Agent Workflows
Pre-launch audit in Cursor:
"Run PageLens on my staging site, list all CRITICAL and HIGH findings, and create GitHub issues for the top 5."
CRO review:
"Fetch the latest scan for example.com, get the CONVERSION quick wins, and summarise what to fix before the campaign launch."
Security sweep:
"List all SECURITY findings with severity HIGH or above from my last scan and show me the evidence for each."
Accepted architecture decision:
"For the CSP unsafe-inline finding, acknowledge this as an intentional Next.js/PPR tradeoff with the rationale from our security docs. Do not mark it as a false positive."
Post-fix validation:
"After I've fixed the findings, start a new scan and compare the score to the previous one."
Pricing
PageLens is pay-per-scan — no subscription required for one-off audits.
| Tier | Price | Pages per scan |
|---|---|---|
| Starter | $1 | 3 pages |
| Professional | $15 | 25 pages |
| Monitor | $5 / month | Weekly automated scans · 5 pages |
Every tier produces the same full report — Starter through Professional differ only by page-count cap, not report depth. The Monitor subscription runs weekly scans automatically and surfaces drift between runs.
Links
- Website: https://www.pagelensai.com
- MCP endpoint:
https://www.pagelensai.com/api/mcp - Dashboard: https://www.pagelensai.com/dashboard
- GitHub: https://github.com/PageLens-AI/pagelensai-mcp-server
- MCP docs: https://www.pagelensai.com/mcp
License
MIT © PageLens AI
Serveurs connexes
Alpha Vantage MCP Server
sponsorAccess financial market data: realtime & historical stock, ETF, options, forex, crypto, commodities, fundamentals, technical indicators, & more
Fetter MCP
Get the most-recent Python package without vulnerabilities, and more.
Gemini MCP
Integrate the full power of Gemini Pro 3 to Claude Code
GroundDocs
A version-aware documentation assistant that connects LLMs to trusted, real-time docs to reduce hallucinations and provide accurate, version-specific responses.
AI Agent Playwright
An AI agent for the Playwright MCP server, enabling automated web testing and interaction.
CursorRules MCP
An intelligent system for managing programming rules, supporting search, versioning, code validation, and prompt enhancement.
xctools
🍎 MCP server for Xcode's xctrace, xcrun, xcodebuild.
Gradio MCP Test
A Python-based MCP server that provides tools to get cat images, either as a direct PNG or a URL for Markdown display.
MCP Image Placeholder Server
Generates placeholder images from various providers like placehold.co and lorem-picsum.
tachibot-mcp
Stop AI Hallucinations Before They Start Run models from OpenAI, Google, Anthropic, xAI, Perplexity, and OpenRouter in parallel. They check each other's work, debate solutions, and catch errors before you see them.
Hex MCP
A server for listing, searching, running, and managing Hex projects.