Search tool calls
search_tool_callsREAD$0.008 / callFIND a recorded call by what is inside it — a page it was about, a word in the answer, an error it returned. Free on every plan. Ranked by WHERE the words were found, not by date: a call whose SUBJECT is /pricing outranks fifty whose answers merely list /pricing among their rows, which is the whole difference between this and grepping a log. Every word must appear somewhere (AND, not OR). The query takes filters inline — tool:get_analytics path:/pricing since:14d failed:true source:cron — and everything else is treated as words. An unknown key:value is searched as two words rather than silently dropped. Reach for it when you know WHAT you are looking for and not when it was taken: 'when did we last audit GEO on the pricing page', 'which reading first showed zero AI citations', 'what failed yesterday'.
Workspace ID (optional when MCP endpoint is auto-scoped). Numeric workspace id — OR the project as the user names it: 'owner/repo', the repo name alone, the site's display name, its docs URL or custom domain. Text is resolved server-side; an ambiguous name returns the candidates instead of guessing, so pass what the user said rather than calling list_workspaces first.
Words to find, plus optional inline filters: tool:, path:, since:, failed:, source:.
Max hits. Default 20.
Call it#
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "search_tool_calls",
"arguments": {
"query": "<query>"
}
}
}curl -X POST 'https://docsbook.io/api/mcp/server' \
-H 'Authorization: Bearer dbk_YOUR_API_KEY' \
-H 'Content-Type: application/json' \
-H 'Accept: application/json, text/event-stream' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"search_tool_calls","arguments":{"query":"<query>"}}}'