Limits
How many calls Athaus accepts, and what happens when there are too many.
Every route has a token bucket: a number of calls at once, then a fixed rate per second. Signed-in callers are counted per account, everyone else per IP address (per /64 network for IPv6).
When there are too many
HTTP/1.1 429 Too Many Requests
content-type: application/json
{ "error": "rate_limited" }There is no Retry-After header. After a 429, wait briefly and double the pause on every further one. On the MCP servers the 429 comes as an HTTP response, not as a JSON-RPC error.
The values
| Route | Rate per second | At once |
|---|---|---|
GET /api/public/properties and GET /api/public/properties/{id} (one shared bucket) | 2 | 10 |
POST /api/public/valuation/location-baseline | 3 | 15 |
| Both MCP servers together, per account | 1 | 30 |
Every write call outside the MCP servers (POST, PUT, PATCH, DELETE) | 1 | 60 |
According to the reference, a page of catalogue results holds at most 60 items (page_size).
Enquiries to providers
An enquiry to a provider (send_inquiry and the form in the exposé) has additional limits so that nobody floods a provider with enquiries:
- at most 5 enquiries in 10 minutes per person,
- at most 5 in 24 hours per email address,
- at most 30 in one hour per listing.
Beyond that the tool answers with rate_limited, and nothing is sent.
Size
A request body may be at most 512 KB; above that the API answers with payload_too_large.