Search Knowledge
The search_knowledge MCP tool performs semantic search across your Knowz vaults. Unlike keyword search, it understands meaning -- searching for 'authentication approach' will find your 'JWT token strategy' document. Filter by vault, tags, date range, and content type to narrow results.
How it works
When an AI agent calls search_knowledge, the query is vectorized and compared against your vault contents using cosine similarity. Results are ranked by relevance and returned with metadata (vault, tags, creation date, author). The search supports natural language queries, boolean filters, and date range constraints. Agents can search specific vaults or across all accessible vaults.
MCP search results showing semantic matches with relevance scores
Screenshot coming soon
Why it matters
AI agents without access to your team's knowledge base operate in a vacuum. They cannot reference past decisions, conventions, or lessons learned. The search tool gives every AI agent in your workflow access to your institutional knowledge. During code review, the agent can search for relevant coding standards. During design, it can find past architectural decisions. During debugging, it can look up known issues and workarounds.
AI agent using search during code review to find relevant standards
Screenshot coming soon