Credits and limits
Understand shared WhatWins AI data credits, result-based charging and MCP rate limits.
MCP and WhatWins AI share one monthly data-credit balance for the organization. Credits measure verified rows returned, not the number of HTTP requests.
How charging works
Before a tool runs, WhatWins reserves the maximum bounded result size. After execution, it settles against the rows actually returned. Failed executions do not charge result credits.
| Example | Maximum result credits |
|---|---|
| 6 ads from one shop | 6 |
| 8 organic posts from one shop | 8 |
| Overview for 2 shops | 2 |
| Discover 5 shops | 5 |
For tools that query several shops, the maximum is generally the requested limit multiplied by the number of shops. The input schema caps both values.
Read current usage
The MCP resource whatwins://usage returns the current shared balance.
{
"jsonrpc": "2.0",
"id": 1,
"method": "resources/read",
"params": { "uri": "whatwins://usage" }
}Successful tool responses also include current usage metadata under _meta.whatwinsUsage.
Rate limits
Rate limits protect the shared service and are enforced per credential, organization and day.
| Plan | Per credential | Per organization | Daily organization budget |
|---|---|---|---|
| Pro | 60/min | 120/min | 2,000 requests |
| Business | 180/min | 360/min | 10,000 requests |
Limits may be tightened temporarily to protect source systems. Do not fire repeated polling requests; ask bounded questions and reuse returned evidence within a conversation.
If credits run out
The server returns JSON-RPC error -32002 with current usage and the required credit estimate. Reduce the number of shops or requested results, or wait for the monthly balance to renew.