For agents and developers

AgentRank is built to be used by AI agents and to be assessed by them. There are two ways in: a deep link that hands a human the prefilled scanner, and the JSON endpoints the web app itself uses.

1. Deep link (works for everyone, no setup)

Send a user to https://jannick.io/?url=example.com to scan a website on their behalf. The scanner is prefilled; anyone not signed in creates a free account (1 free scan, no credit card), after which the scan starts automatically.

2. JSON endpoints (signed-in session required)

These are the functions the app calls. They are authenticated with the user's AgentRank session, the same session the browser holds after login, so there is no separate API key to manage or leak.

Run a scan

Endpoint: POST https://jannick.io/api/apps/6a9d111004ca350a0146906d/functions/runAudit
Body: { "url": string, "force"?: boolean } — interactive scans already read the site live; force is a marker for paid plans. See the OpenAPI spec for optional save and follow-up input.
Response (200): { "result": { site_name, overall_score, grade, summary, categories, top_fixes, generated_llms_txt, generated_schema, detected_cms, live_crawled }, "saved": boolean, "auditId": string|null }

Read a shared report

Endpoint: POST https://jannick.io/api/apps/6a9d111004ca350a0146906d/functions/getPublicReport · Body: { "id": string } (24-hex report id).Only reports whose owner turned on 'share a public link' are returned. Agency reports from an active subscription can also be opened without an AgentRank login (white-label share link). The same report can be viewed at https://jannick.io/r/<reportId>.

Benchmark sites against competitors

Endpoint: POST https://jannick.io/api/apps/6a9d111004ca350a0146906d/functions/getBenchmark
Body: { "urls": ["your-site.com", "competitor-a.com", "competitor-b.com"] } — 2 to 3 URLs; the first is your site, the rest are competitors.
Response (200): { "rows": [{ url, host, available, auditId, site_name, overall_score, grade, categories, scanned_at, public }], "insights": [{ host, signal, delta }], "generated_at": string }
Reads the most recent audit per host from records you can see (your own, plus public audits as a fallback) — no new scans are started, so it doesn't count against your fair-use limit. insights lists every signal where a competitor scores 5 points or more above your site. Also available to Pro and Agency users on the comparison page.

Limits

• Free: 1 scan. Pro: max 50 per day and 200 per 30 days. Agency: 100 per billing period and max 50 per day. Every call counts, including with save:false.
• Only public http(s) websites on standard ports can be scanned; private or internal addresses are refused.
• A scan takes about a minute: it reads the homepage, robots.txt, sitemap.xml and llms.txt live and assesses six categories.

Error codes

• 400 invalid_url — not a public website address.
• 401 auth_required — no signed-in session.
• 402 limit_reached — free scans used up; upgrade to Pro.
• 402 agency_quota — agency subscription monthly quota reached.
• 429 fair_use — paid plan limit reached (50 scans per 24 hours / 200 per 30 days). scope is day or month; wait at least the number of seconds in the Retry-After header before trying again. The other window may still apply.
• 502 scan_failed — the site couldn't be read well enough to assess.

Machine-readable manifests

• /openapi.json — OpenAPI 3.0 spec for the endpoints above (request/response schemas, error codes).
• /.well-known/agent-card.json — capability card with AgentRank's HTTP skills, endpoints and authentication; not an A2A protocol server.
• /.well-known/mcp.json — MCP-style manifest of the same tools (no hosted MCP server yet; points to OpenAPI + agent card).
• /agents.json — registry of the agents we operate (the AgentRank/1.0 scanner) and how to opt out.
• /llms.txt — plain-text overview for LLMs; /robots.txt contains our Content-Signal (search, AI input and training allowed).
• /changelog.txt — date-stamped API and pricing changes for agents, with links to the latest OpenAPI spec.

Programmatic access at scale

Agencies: see the Agency plan for client scans and white-label reports. For integrations and MCP questions, email leon@jannick.io.. A machine-readable overview of AgentRank itself is at /llms.txt. Scans only read public pages, see the privacy policy.