What problem does it solve? Building AI agents that remember conversations, call tools, and stream responses requires stitching together state management, vector search, and real-time infrastructure. This Skill provides ready-to-use patterns for building stateful agents on Convex so conversation history, embeddings, and workflows persist without extra infrastructure. ## Core Features & Use Cases - Thread & Message Management: Create conversation threads, list user threads, and retrieve message history stored in Convex tables. - Tool Integration: Define Convex functions as agent tools (knowledge search, task creation, external API calls) that the LLM can invoke during chat. - RAG & Vector Search: Generate embeddings on document insert and query them with vector indexes for retrieval-augmented generation. - Workflow Orchestration: Run multi-step agent workflows (search, analyze, summarize) with status tracking in the database. - Use Case: Add an AI support assistant to your app that remembers each customer's conversation, searches your knowledge base for answers, and creates follow-up tasks automatically. ## Quick Start Ask the AI to set up a Convex agent with a chat thread, a knowledge search tool, and streaming responses using the @convex-dev/agent component.
