1. Get a key

  1. Open Settings > API keys

    In the portal, open Settings, then API keys.

  2. Create a key and name it

    Name it after where it will be used ("Reporting sheet", "Client dashboard"), so it's clear later which one to turn off. Only account admins can create keys; everyone on the team can see the list.

  3. Copy it straight away

    The full key is shown once, right after it's made. Paste it into the tool that needs it. If it gets lost, revoke it and make a new one.

Treat a key like a password

Anyone holding the key can read this account's data. Don't email it or post it anywhere public. If a key may have leaked, open Settings > API keys and click Revoke: it stops working at once.

2. Ask for data

Most tools that "connect to an API" ask for two things: a web address and a key. The key goes in a header called Authorization, written as Bearer followed by the key. Here is the same request as a command, for whoever sets the tool up:

curl -H "Authorization: Bearer crk_your_key_here" \
  https://api.crawlready.ai/api/v1/scans

Every answer comes back as JSON, the plain data format spreadsheets and dashboards read.

3. What can be read

AddressWhat comes back
/api/v1/Which account the key belongs to, and this list.
/api/v1/sitesThe account's sites: whether each is being monitored, its latest grade and its next check.
/api/v1/scansEvery scan with its date, score and grade, newest first.
/api/v1/scans/{job_id}One scan in full: score, grade and the whole report with its findings.
/api/v1/monitoringThe monitoring plan: sites monitored of the plan's limit, paused sites, and the current period.
/api/v1/reportsThe monthly monitoring reports, newest first.
/api/v1/reports/{site_id}/{month}One monthly report, with the month written like 2026-09.

4. Limits

  • 60 requests a minute per key. Each answer says how many are left that minute. Going over returns error 429 with how many seconds to wait.
  • Lists come 50 at a time (ask for up to 100 with ?limit=100). When there are more, the answer includes next_before; add ?before= with that number to get the next batch.
  • Read only. Anything other than reading is refused (error 405). Scans, monitoring and team changes stay in the portal.
  • Only this account. A key reads the data of the account it was made in, and nothing else.

5. Ask an AI assistant instead (MCP)

AI assistants such as Claude and Cursor can use the same key to look things up for you — ask "which site's score dropped since last month?" or "summarize the latest scan" and the assistant reads the answer from this account. It uses MCP, the standard way AI assistants connect to tools. It's the same key, the same read-only access to this one account, and the same 60 requests a minute.

The address to give the assistant is https://api.crawlready.ai/mcp, with the key sent as Authorization: Bearer followed by the key.

  • Claude Code: run this once (whoever sets it up can paste it into a terminal): 
    claude mcp add --transport http ai-readiness https://api.crawlready.ai/mcp \
  --header "Authorization: Bearer crk_your_key_here"
  • Cursor: open Settings > MCP > Add a new server, or add this to the mcp.json file: 
    {
  "mcpServers": {
    "ai-readiness": {
      "url": "https://api.crawlready.ai/mcp",
      "headers": { "Authorization": "Bearer crk_your_key_here" }
    }
  }
}
  • Other AI apps: any app that can add an MCP server by web address and a header works the same way: the address above, and the key in the Authorization header.

The assistant gets seven tools, all read-only: the account, its sites, the list of scans with scores, one scan's full report, the monitoring plan, the monthly reports, and one monthly report.

Error 401?

The key is missing, mistyped or revoked. Check it starts with crk_ and is sent as Authorization: Bearer followed by the key, or make a new one.