Model Context Protocol
Exposing the API to an AI agent as tools.
The API publishes a tool manifest so an agent can call it directly rather than through hand-written wrappers.
curl https://api.registercheck.de/toolsEvery operation in the API reference is listed as a tool, with its method, path and parameters.
{
"tools": [
{
"name": "search_companies_search_companies_get",
"description": "Search Companies",
"method": "GET",
"path": "/search/companies"
}
]
}A plugin manifest is served at /.well-known/ai-plugin.json.
Invoking a tool
curl -X POST https://api.registercheck.de/invoke \
-H "Authorization: Bearer $REGISTERCHECK_API_KEY" \
-H "Content-Type: application/json" \
-d '{"tool_name": "search_companies_search_companies_get",
"arguments": {"name": "Enduro", "limit": 1}}'Tool calls are billed exactly like the endpoint they wrap, and the same credit costs apply.
Tool invocation is currently not working
/tools lists the manifest correctly, but POST /invoke rejects authenticated tools
with 401 {"detail": "Valid Bearer token required in Authorization header"} even when
the same key succeeds against the REST endpoint, and unauthenticated tools resolve to
the wrong internal path and return an inner 404. Until this is fixed, point your agent
at the REST endpoints directly — an OpenAPI-driven tool layer over
https://api.registercheck.de/v1 works today.
Giving an agent the docs
Every page here is available as Markdown for an agent to read:
https://docs.registercheck.de/llms.txt— the page indexhttps://docs.registercheck.de/llms-full.txt— the whole documentation as one file- any page URL with
.mdappended — that page as Markdown