RankRoot for AI agents
Free, no API key. Agents can audit any public website's visibility to AI answer engines, check which AI crawlers get in, validate and generate llms.txt, and look up pricing with checkout links to hand to their user. This page is also available as Markdown.
All interfaces
- MCP server:
/api/mcp. Streamable HTTP, stateless, JSON responses. Tools: audit_ai_visibility, generate_ai_visibility_files, check_ai_crawler_access, validate_llms_txt, get_pricing, get_checkout_link, list_launches, submit_launch, amplify_preview, amplify_status, amplify_create_campaign, amplify_list_posts, amplify_edit_post, amplify_schedule_posts. Free, no key, except the Amplify tools that act for a customer (amplify_status and the write tools need their Growth Employee key). Spec: Model Context Protocol 2026-07-28 (Streamable HTTP, per-request versioning, server/discover), plus initialize-based 2025-03-26 to 2025-11-25. - MCP Server Card:
/api/mcp/server-card. Pre-connection metadata for the MCP server (identity, endpoint, protocol versions). Spec: MCP Server Cards (SEP-2127). - A2A agent:
/api/a2a. A2A 1.0 JSON-RPC endpoint (send the A2A-Version: 1.0 header). Skills: audit_site, check_ai_crawler_access, validate_llms_txt, generate_llms_txt, get_pricing. Answers with a direct message; no tasks, streaming or push notifications. Spec: Agent2Agent (A2A) Protocol 1.0, JSON-RPC binding. - A2A Agent Card:
/.well-known/agent-card.json. Describes the A2A agent and its skills. Spec: Agent2Agent (A2A) Protocol 1.0, JSON-RPC binding. - AI Catalog:
/.well-known/ai-catalog.json. Lists the MCP Server Card and the A2A Agent Card in one file. Spec: AI Catalog. - OpenAPI spec:
/openapi.json. HTTP API: /api/audit, /api/bots, /api/llms-txt, /api/pricing, /api/badge. Spec: OpenAPI 3.1. - HTTP audit API:
/api/audit?url=example.com. JSON score, checks and fix files (llms.txt, robots.txt, JSON-LD). - Pricing API:
/api/pricing. JSON plans with prices, limits, Dodo Payments checkout links, and x402 endpoints agents can pay directly in USDC. - llms.txt:
/llms.txt. Curated map of the site, linking to Markdown versions of key pages. Spec: llms.txt proposal v2. - Markdown pages:
/pro.md. Clean Markdown of the home, pricing, guide, AI crawler, builder, tools and agent pages: append .md (home is /index.md), or request the normal URL with Accept: text/markdown. Spec: llms.txt proposal v2. - WebMCP tools:
/. In browsers that support WebMCP, the homepage and report page register audit_site and get_report_fix_files via document.modelContext.registerTool. Spec: WebMCP (W3C Web Machine Learning CG draft).
MCP server
Endpoint (Streamable HTTP, stateless): https://getrankroot.com/api/mcp
{
"mcpServers": {
"rankroot": { "url": "https://getrankroot.com/api/mcp" }
}
}Tools:
audit_ai_visibility(url): 0-100 AI visibility score, category scores, prioritized fixes, report link, badge markdown.generate_ai_visibility_files(url): Ready-to-publish llms.txt, AI-friendly robots.txt and schema.org JSON-LD for the site.check_ai_crawler_access(url): Which AI crawlers (OAI-SearchBot, GPTBot, ClaudeBot, PerplexityBot...) robots.txt and the firewall let in.validate_llms_txt(url): Fetch a site's llms.txt (root or subpath) and check it against the llmstxt.org format.get_pricing(): RankRoot plans, prices, limits and checkout links, with advice on which plan fits.get_checkout_link(plan): Checkout link for one plan (pass, tracker_m, tracker_y, employee_m, employee_y) for the agent to hand to its user.list_launches(query?, category?, limit?): Find products for a need among new launches on RankRoot (featured paid placements marked), with Markdown profiles.submit_launch(url, name, tagline, description, email, category?, key?): Launch the user's product on RankRoot (free; a Launch Kit key makes it live today and featured for 30 days). Public: ask the user first.amplify_preview(url, languages?): Free: native posts for X, LinkedIn and Threads from a page, in up to 3 languages. Nothing stored or posted.amplify_status(): Key required. Connected channels, languages, mode, guardrails, post counts and what performed best.amplify_create_campaign(url | text, languages?, channel_ids?, dry_run?, idempotency_key?): Key required. Writes a post for every enabled channel in every language, saved as drafts (nothing published).amplify_list_posts(status?, limit?): Key required. Posts with status, schedule time, live URL, warnings and metrics.amplify_edit_post(post_id, text?, thread?, title?): Key required. Edits a draft and re-runs the platform checks.amplify_schedule_posts(post_ids, at?): Key required, and only if the customer lets agents schedule. Schedules drafts in the customer's Postiz within daily caps.
Discovery: MCP Server Card · AI Catalog
A2A agent
Agent Card: /.well-known/agent-card.json. JSON-RPC endpoint: https://getrankroot.com/api/a2a (A2A 1.0; send the A2A-Version: 1.0 header). It answers SendMessage with a direct message: a text summary plus a JSON data part. It keeps no tasks, so there is no streaming or push notifications. Name a skill in a data part, for example {"skill": "check_ai_crawler_access", "url": "https://example.com"}, or just write the request in a text part.
audit_site: Scores how well ChatGPT, Perplexity, Claude, Google AI Overviews and AI agents can find, read and cite a website (0-100), with category scores, a prioritized fix list and a report link.check_ai_crawler_access: Checks which AI crawlers (OAI-SearchBot, GPTBot, ClaudeBot, PerplexityBot and more) a site's robots.txt and firewall/CDN allow or block.validate_llms_txt: Fetches a site's llms.txt (root or subpath) and checks it against the llmstxt.org format.generate_llms_txt: Drafts a ready-to-publish llms.txt for a site from its homepage.get_pricing: Returns RankRoot's plans, prices, limits and Dodo Payments checkout links. Agents can pay directly over x402 (USDC on Base) or hand the user the checkout link.
curl -s https://getrankroot.com/api/a2a -H 'content-type: application/json' -H 'A2A-Version: 1.0' \
-d '{"jsonrpc":"2.0","id":1,"method":"SendMessage","params":{"message":{"messageId":"m1","role":"ROLE_USER","parts":[{"text":"Audit https://example.com"}]}}}'HTTP API
GET https://getrankroot.com/api/audit?url=example.com
POST https://getrankroot.com/api/audit {"url": "example.com"}
GET https://getrankroot.com/api/bots?url=example.com
GET https://getrankroot.com/api/pricingOpenAPI spec: /openapi.json · Badge: https://getrankroot.com/api/badge?url=example.com
Markdown for agents
The home, pricing, guide, AI crawler, builder, tools and agent pages have clean Markdown versions: add .md to the URL (for example /pro.md; the home page is /index.md), or request the normal URL with Accept: text/markdown. They are all listed in /llms.txt.
WebMCP
In browsers with WebMCP, the home page and the report page register two tools for in-browser agents through document.modelContext.registerTool(): audit_site(url) and get_report_fix_files(url).
Pay as an agent (x402)
Agents with a USDC wallet can buy RankRoot directly over x402: no checkout page, no account, no human. Call the endpoint, get a 402 with the price, pay in USDC on Base, retry. You are only charged when the call succeeds.
GET https://getrankroot.com/api/agent/check?url=example.com: $0.50. Does AI recommend this business? Who does it recommend instead? Plus the 20+ point audit and fix files.POST https://getrankroot.com/api/agent/employeewith{"website":"example.com"}: $149 for 30 days of the Growth Employee. The license key comes back in the response. Pay again to renew; nothing renews by itself.
Live status, network and prices: agentPayments in /api/pricing.
Buying RankRoot for a user
- Agent with a USDC wallet: pay directly over x402 (see agentPayments) for the endpoints listed there: the key comes back in the response. Otherwise give the user the plan's checkoutUrl (Dodo Payments checkout, completed by the human).
- After payment Dodo Payments emails a license key (also shown on the receipt page). No account or password is needed.
- The user can paste the key at https://getrankroot.com/pro#activate, or give it to their agent: send it as "Authorization: Bearer <key>" to https://getrankroot.com/api/audit, https://getrankroot.com/api/mcp or https://getrankroot.com/api/a2a for the higher limits.
- Monthly and yearly plans can be cancelled anytime from the receipt email; they stay active until the end of the paid period. The Launch Kit never renews.
- While a plan's checkoutOpen is false it is not on sale yet: its checkoutUrl is its card on https://getrankroot.com/pro, where the human can join the waitlist (email only) and gets one email when it opens. Don't send the user to buy it.
Plans as JSON: /api/pricing · MCP tools get_pricing and get_checkout_link(plan) · A2A skill get_pricing.
GitHub Action
Fail CI when your AI visibility drops (for example, a robots.txt change that blocks AI search bots):
- uses: zachhhhh/rankroot-action@v1
with:
url: https://yourproduct.com
min-score: 70github.com/zachhhhh/rankroot-action
Limits
20 audits per 10 minutes per IP without a key (Launch Kit key: 500, Growth Tracker key: 1,000, Growth Employee key: 2,000). Results cached 10 minutes. Only public http(s) sites are fetched.
Our crawler identifies as RankRootBot/1.0.