Landing page audits for AI agents: MCP server and API

PageCritic's audit is available to agents as a remote MCP server and as a plain REST API. Ask Claude, Cursor or your own agent to "audit my landing page" and it gets the same result a visitor gets on the homepage: a conversion score out of 100, 7 sub-scores, the top 3 issues with fixes, what already works and an AI-readiness score. No sign-up needed for the free result. An optional API key raises the limits, and credits return the full report.

MCP server

URL: https://pagecritic.app/mcp (Streamable HTTP). Listed in the official MCP Registry as app.pagecritic/pagecritic.

audit_landing_page(url): loads the page in a real browser (desktop and mobile) and returns the score, sub-scores, top 3 issues with fixes and what works. Takes 30–90 seconds.
check_ai_readiness(url): can GPTBot, ClaudeBot, PerplexityBot and other AI crawlers read the page? Blocked bots, robots.txt rules, text without JavaScript, JSON-LD, sitemap, llms.txt. About 5–15 seconds.
get_audit_result(audit_id): fetches an audit that was still running.

Claude Code

claude mcp add --transport http pagecritic https://pagecritic.app/mcp

Claude (web and desktop): Settings → Connectors → Add custom connector, and paste https://pagecritic.app/mcp.

Cursor (~/.cursor/mcp.json)

{ "mcpServers": { "pagecritic": { "url": "https://pagecritic.app/mcp" } } }

VS Code (.vscode/mcp.json)

{ "servers": { "pagecritic": { "type": "http", "url": "https://pagecritic.app/mcp" } } }

REST API

OpenAPI 3.1 spec: /openapi.json.

curl -s https://pagecritic.app/api/v1/audit \
  -H 'Content-Type: application/json' -d '{"url":"https://example.com"}'

curl -s 'https://pagecritic.app/api/v1/readiness?url=https://example.com'

curl -s https://pagecritic.app/api/v1/audit/<audit_id>

POST /api/v1/audit waits up to 85 seconds. If the result still says "status": "running", fetch GET /api/v1/audit/{audit_id} a little later. Send "wait": false to get the id back at once.

API keys and credits (optional)

A key is free and needs no card. With a key you get 20 audits per hour instead of 5. Buy credits, and each audit made with the key returns the full report in full_report: 10–15 ranked issues, headline and CTA rewrites, quick wins, the mobile review and the AI-readiness fixes. 10 full reports cost $49 (one-time, no subscription). Send the key as Authorization: Bearer <key> to the REST API or the MCP server.

# 1. create a key (shown once)
curl -s -X POST https://pagecritic.app/api/v1/keys

# 2. get a Stripe Checkout link for 10 credits; a person opens it and pays
curl -s -X POST https://pagecritic.app/api/v1/credits -H 'Authorization: Bearer pc_...'

# 3. audit; Idempotency-Key makes retries return the same audit without spending another credit
curl -s https://pagecritic.app/api/v1/audit -H 'Authorization: Bearer pc_...' \
  -H 'Idempotency-Key: my-run-42' -H 'Content-Type: application/json' -d '{"url":"https://example.com"}'

# balance
curl -s https://pagecritic.app/api/v1/keys/me -H 'Authorization: Bearer pc_...'

Spend cap: at most 25 credits per key per day; past that, audits return the free result. A failed audit gives the credit back. In Claude Code: claude mcp add --transport http pagecritic https://pagecritic.app/mcp --header "Authorization: Bearer pc_..."

Pricing and limits

Free: the full free result through either interface. 5 audits and 15 readiness checks per hour per client (20 audits per hour with a free key). Agents asking for a URL audited in the last 24 hours get that same result.
API credits: 10 full reports for $49, see above.
Every result includes a report_url. Opening it shows the visual report, where a person can unlock the full report for $19 one-time per page: 10–15 more ranked issues, 3 headline and 3 CTA rewrites, 5 quick wins, the mobile review and copy-paste AI-readiness fixes. 14-day refund guarantee.
Machine-readable pricing: /pricing.json. Need higher limits? Email hello@pagecritic.app.

Only audit pages you own or are allowed to review. Results are AI-generated opinions, not guarantees (see terms).