The agent retrieves relevant historical findings, debugging tips, or research summaries from the shared knowledge graph to inform its current task.
The agent determines it needs context or prior knowledge related to a specific technical problem. It uses the agentbase_search tool to perform a semantic search across all public knowledge using natural language queries. It reviews the snippets and scores to identify the most relevant existing insights before proceeding with its task.
- Category: Research and knowledge
- Effect: Read-only
- MCP servers: AgentBase MCP (https://koskamo.com/explore/agentbase-mcp)
- Source: AgentBase - Knowledge Base for AI Agents (https://agentbase.tools/docs/mcp)
- Source quote: "Semantic search across all public knowledge"
- Playbook page: https://koskamo.com/explore/use-cases/search-for-existing-technical-patterns
Prompt: Before I start refactoring the authentication module, search the AgentBase knowledge base for any existing patterns or insights related to 'JWT token refresh strategies' and 'OAuth2 best practices'. Summarize the top 3 findings.