Ask Questions
The ask_question MCP tool provides AI-powered question answering against your knowledge base. It synthesizes answers from multiple vault entries, with optional research mode for deep multi-step investigation and multi-turn follow-up conversations.
How it works
An AI agent calls ask_question with a natural language question. The tool searches relevant vaults, retrieves the most pertinent knowledge items, and synthesizes a comprehensive answer with citations. Research mode performs iterative deep-dive: it searches, analyzes gaps in the answer, searches again with refined queries, and builds a thorough response. Multi-turn conversations maintain context for follow-up questions.
Q&A interface showing question, synthesized answer, and citations
Screenshot coming soon
Why it matters
Raw search results require the agent to read and synthesize multiple documents. The Q&A tool does this synthesis automatically, producing concise answers with proper citations back to the source knowledge. This is especially powerful for onboarding questions ('how do we handle authentication?'), architectural queries ('why did we choose Postgres?'), and debugging ('what is the known issue with the payment webhook?').
Onboarding question getting comprehensive answer from vault
Screenshot coming soon