add_knowledge_chapter

writes

One of the tools an AI assistant can call on an Avrosh account.

What it does

Write one NEW chapter of knowledge: a title and a body of plain text, in the operator's own words - opening hours, what is and is not offered, policies, anything a customer would phone to ask. Only write what the human actually told you; never invent a fact about the business. Call list_knowledge first: if a chapter on the subject already exists this tool refuses with its id, so you update it instead of creating a duplicate that argues with it.

This is the exact text the model reads before deciding to call.

Parameters

business_idstringrequired

The id returned by list_businesses.

titlestringrequired

The chapter heading, e.g. 'Opening hours'. Max 200 chars.

bodystringrequired

The facts as plain text, written the way the business would say them. Line breaks are kept.

Response

The result arrives as an MCP content block; the JSON in the panel is what the text field parses to.

Connecting

Point any MCP client at https://us.avrosh.com/mcp and log in — OAuth 2.1, no key to paste.

The panel here uses the other door: a bearer key made in the dashboard, for a script or a cron job with no browser to sign in with. Both reach the same endpoint. Full notes on the reference overview.

curl -X POST https://us.avrosh.com/mcp \
  -H "Authorization: Bearer $AVROSH_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "add_knowledge_chapter",
      "arguments": {
        "business_id": "926135ac-487d-4367-8bad-25397d0d4b87",
        "title": "Opening hours",
        "body": "Mon-Sat 09:00-18:00, closed Sunday."
      }
    }
  }'