Site Auditor · MCP-installable agent
Does ChatGPT recommendyour site or your competitor's?
Site Auditor scores any public website on AI-search readiness in ~20 seconds. Get a 0–100 score, prioritized findings, and the exact fixes to rank in ChatGPT, Perplexity, Google AI Overview, and Claude.
Free tier: 1 audit / calendar month · No account needed · Pro from $29/mo
The problem
Your site was built for Google blue links.
It's being graded by ChatGPT.
When a shopper asks “what's the most comfortable lightweight summer shoe?” in ChatGPT — does Crocs show up? When someone asks Perplexity for “the best running shoes for marathon training” — is Academy Sports in the answer? When a buyer asks Claude “where should I buy an engagement ring online?” — does Signet appear with a citation?
The answer depends on whether your site has the metadata, structured data, and AI-crawler signals that LLM-powered search engines actually parse. Most retailers built their sites for human SEO (Google blue links). They've never been audited for AI citation readiness.
By the time you notice your share of voice dropping in AI answers, your competitors are already ranking. Site Auditor closes that gap.
What you get per audit
Score, findings, and the fixes that move the needle
0–100 AI-readiness score
One number, instantly comparable across audits and across competitors.
Detected ecommerce platform
Shopify, Salesforce Commerce Cloud, BigCommerce, Adobe Commerce, custom — auto-identified.
Prioritized findings
HIGH / MEDIUM / LOW — exactly what to fix and why it matters for AI search.
Strengths the site already has
What's working — so you know which optimizations to protect.
AI crawler directives
/robots.txt review for GPTBot, ClaudeBot, Google-Extended, PerplexityBot.
Structured-data review
JSON-LD: Product, Organization, WebSite, BreadcrumbList — present and well-formed?
/llms.txt detection
The emerging standard for AI manifests — published or missing?
Powered by Claude reasoning
Findings come with a one-sentence diagnosis you can paste into a ticket.
MCP-native
Runs in any MCP client: Cursor, Claude Code, Claude Desktop, Cline, Windsurf, GitHub Copilot, Antigravity.
Who it's for
Built for retail and ecommerce ops
The teams who care most about AI-citation readiness — because their categories are where AI assistants are already shifting buyer behavior.
Crocs
crocs.com
“What's the most comfortable lightweight summer shoe?”
Audit reveals whether product schema includes aggregateRating, availability, and image fields that AI shopping cards require.
Academy Sports
academy.com
“Best running shoes for marathon training.”
Checks whether GPTBot and ClaudeBot have read access for product crawls, and whether /llms.txt declares the brand's AI policy.
Signet Jewelers
signetjewelers.com
“Where should I buy an engagement ring online?”
Verifies Organization JSON-LD (logo, sameAs, contactPoint) so AI assistants identify the brand authoritatively in answers.
Also useful for
Pre-launch audits before paid traffic spends · Quarterly competitive scans across your category · Founder + indie-hacker site validation · Agency client deliverables · SEO + content team measurement of AI search alongside Google
Install in 30 seconds
Drop it into your MCP client
No account required for the free tier. Pro customers paste their key in the optional apiKey field and the same install upgrades automatically.
Claude Code (terminal)
shell
claude mcp add --transport http site-auditor https://multipliers.ai/mcp/site-auditor
Add --header "X-API-Key: mp_live_…" at the end if you have a Pro key.
Cursor / Cline / Windsurf
json
{
"mcpServers": {
"site-auditor": {
"url": "https://multipliers.ai/mcp/site-auditor"
}
}
}Settings → MCP → Add new MCP server → paste.
Claude Desktop
json
{
"mcpServers": {
"site-auditor": {
"url": "https://multipliers.ai/mcp/site-auditor"
}
}
}Edit ~/Library/Application Support/Claude/claude_desktop_config.json and merge under `mcpServers`. Restart Claude.
GitHub Copilot (VS Code 1.95+)
json
{
"servers": {
"site-auditor": {
"type": "http",
"url": "https://multipliers.ai/mcp/site-auditor"
}
}
}Save as `.vscode/mcp.json` in your workspace. Requires Copilot Chat MCP support.
Google Antigravity
json
{
"mcpServers": {
"site-auditor": {
"url": "https://multipliers.ai/mcp/site-auditor"
}
}
}Settings → MCP servers → Add server → paste.
Then ask in your MCP client:
“Audit https://example.com for AI readiness.”
Slack / Teams / Discord: not native MCP clients. Wrap Site Auditor in a custom bot using the same HTTPS endpoint with X-API-Key header. We provide a reference Slack bot template — docs/integrations/slack.
Plans
Start free. Upgrade when you outgrow it.
Free
1 audit / calendar month
- No account required
- Install via Smithery in one click
- Same audit engine as Pro
- Identity-keyed rate limit (per session, not per IP)
- Resets on the 1st of every month
Pro
25 audits / month
- All depth from free tier (21 AI crawlers, schema completeness, page hygiene, score breakdown)
- Shareable scorecard URLs (multipliers.ai/scorecards/<id>)
- PDF export — multi-page A4, brand-styled
- 30-day retention on shared scorecards
- Priority queue (no shared free-tier limit)
- Email + chat support
Pro Unlimited
Unlimited audits
- Everything in Pro
- No per-period cap
- Best for agencies + high-volume teams
- Bring your own AI key (BYOK) optional
All charges appear on your statement as MULTIPLIERS.AI.
Under the hood
How Site Auditor works
- 1
Fetch + parse
The auditor fetches the HTML of the target URL with a real browser-like User-Agent. It parses meta tags, link relations, robots.txt, /llms.txt, JSON-LD blocks, and OpenGraph data — no third-party scrapers.
- 2
Detect platform
Heuristics identify the underlying ecommerce platform (Shopify, SFCC, BigCommerce, Adobe Commerce, custom). Knowing the platform shapes the recommendations — Shopify needs different schema fixes than SFCC.
- 3
Score against AI-search criteria
Each signal contributes to the 0–100 score: structured data presence + correctness, AI-crawler directives, /llms.txt, page metadata, internal linking patterns, mobile-readiness for AI rendering.
- 4
Reason with Claude
Findings are passed through Claude (Sonnet 4.6) which writes the human-language diagnosis: “Add Product JSON-LD with aggregateRating to compete in AI shopping cards. Required for ChatGPT shopping results.”
- 5
Return as MCP tool result
The audit is returned as a markdown response your MCP client renders inline. Your AI agent can then take action — file a ticket, update a doc, or just summarize for the user.
Privacy + security
What we collect, what we don't
What we DO collect
- The URL you submit for audit (logged for support)
- The publicly-fetchable HTML of that URL (parsed, not stored)
- Anonymized usage counters for rate-limiting
- API key hash (never plaintext) for Pro tier authentication
What we do NOT collect
- Anything behind authentication on your site
- Personal data of your visitors
- The full HTML payload after analysis (discarded)
- Anything we can't justify needing for the audit
See our Privacy Policy and Terms. Cancel anytime via Manage subscription — Stripe Customer Portal handles refunds + downgrades.
FAQ
Common questions
Do I need an account to use the free tier?▾
No. Install via Smithery and start auditing — your MCP client gets a per-session identity that we use to track your 3-per-hour / 10-per-day limit.
How does the Pro tier upgrade work?▾
Subscribe at /pricing, get an API key on the post-checkout page, paste it in the optional `apiKey` field at install time on Smithery (or pass `X-API-Key` header directly). No reinstall — same MCP server, just authenticated.
Can I audit a site behind authentication?▾
Not in v1. Site Auditor only works with publicly-fetchable URLs. Authenticated audits are on the roadmap — talk to sales if you need it for an enterprise contract.
How accurate is the score?▾
The 0–100 score reflects what AI search engines look for: structured data, AI-crawler directives, /llms.txt, page metadata. It's calibrated against the major AI assistants' published parsing logic. It's a directional score — use the prioritized findings as your action list.
Can I share an audit with my team?▾
Pro tier returns a shareable scorecard URL with each audit. Free tier is per-session only.
What MCP clients are supported?▾
Any client that speaks MCP Streamable HTTP (spec 2025-03-26): Cursor, Claude Code, Claude Desktop, Cline, Windsurf, GitHub Copilot, Google Antigravity, plus any custom client. Slack/Teams/Discord aren't native MCP — wrap with a bot.
What happens to my key if I cancel my subscription?▾
Stripe sends us a webhook on cancellation; we automatically revoke the API key tied to that subscription. Free tier still works without a key.
Is this part of a larger platform?▾
Yes — Site Auditor is one of nine production agents in the Multipliers catalog. See /pricing for the full agent list, or talk to sales about the Enterprise plan (full agent suite + governance + observability + your AI runtime).
Find out what AI sees when it looks at your site
Free to try, no account needed. Install in 30 seconds and run your first audit in your favorite MCP client.