convex-agents

Build persistent AI agent systems with threads, tools, and vector search in Convex.

Updated May 4, 2024

Search Tags:#ai #rag #agents #streaming #tools #workflows #convex

Ready to run

npx skills add https://github.com/kcrlee/dots --skill convex-agents-kcrlee

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: convex-agents
Source: https://github.com/kcrlee/dots/tree/main/claude/.claude/skills/convex-agents
Command: npx skills add https://github.com/kcrlee/dots --skill convex-agents-kcrlee

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps developers build persistent, stateful AI agents in Convex without reinventing thread storage, streaming, tool execution, or retrieval logic.

Core Features & Use Cases

Thread management: Create, list, and fetch conversation history that survives restarts.
Tool integration and streaming: Wire Convex functions and external APIs into agent chats while delivering token-by-token responses.
RAG and workflows: Add embeddings, vector search, knowledge retrieval, and multi-step orchestration for research or assistant workflows.
Use case: Use it to implement a customer support agent that remembers prior turns, searches your knowledge base, creates tasks, and streams replies in real time.

Quick Start

Ask for a Convex agent implementation that sets up persistent threads, tool calls, streaming chat, and retrieval augmented search for your app.

Frequently Asked Questions about convex-agents

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I build persistent AI agents in Convex?▼

Persistent AI agents in Convex are built using thread storage, tool execution, streaming, and retrieval logic to maintain state across restarts without reinventing the underlying infrastructure.

Can I stream token-by-token responses from a Convex agent?▼

Yes, streaming token-by-token responses from a Convex agent is supported through built-in streaming message delivery, allowing real-time chat experiences while executing tool calls.

What's the best way to add retrieval-augmented generation to a Convex chat application?▼

Adding retrieval-augmented generation to a Convex chat application involves using embeddings and vector search to retrieve knowledge base content, feeding it into the agent workflow for context-aware responses.

Does Convex support durable workflow handling for long-running agent orchestration?▼

Convex supports durable workflow handling for long-running agent orchestration, ensuring multi-step research and assistant workflows persist reliably through tool invocation and state management.

How do I wire external APIs and tool calls into a Convex AI assistant?▼

External APIs and tool calls are wired into a Convex AI assistant by integrating Convex functions with tool execution logic, enabling the agent to invoke external services during conversation streaming.

Why do my AI agent conversations lose context after server restarts?▼

AI agent conversations lose context after restarts without persistent thread storage; implementing thread management ensures conversation history survives, allowing agents to remember prior turns reliably.