AI Agent Tools List 2026: 30+ Tools Categorized for Developers and Builders

A categorized list of 30+ AI agent tools: frameworks, memory, retrieval, browser automation, and MCP.

## AI Agent Tools List 2026 The AI agent stack has matured into six layers. Here is the working list, with star counts from the GitHub API snapshot of 2026-07-31. ### 1. Agent frameworks - LangChain / LangGraph - 143,097 stars - AutoGen - 60,129 stars - CrewAI - 56,432 stars - LlamaIndex (agents and RAG) - 51,260 stars - Semantic Kernel - 28,396 stars - Haystack - 26,073 stars ### 2. Memory - mem0 - 62,202 stars - Zep - graph memory with automatic entity extraction - Letta (formerly MemGPT) - self-editing memory ### 3. Retrieval / RAG - Qdrant - 33,697 stars - Milvus - 45,439 stars - pgvector - 22,417 stars - Weaviate - 16,677 stars ### 4. Browser automation - browser-use - 107,406 stars - Playwright MCP - 5,620 stars - Skyvern - AI-native browser automation for forms ### 5. MCP (Model Context Protocol) - modelcontextprotocol/servers - 89,093 stars - awesome-mcp-servers - 91,652 stars (10,000+ servers) - mcp python-sdk - 23,816 stars ### 6. Orchestration and platforms - n8n - 198,849 stars - Dify - 150,912 stars - Langflow - 152,678 stars - OpenWebUI - 147,464 stars ### 7. Observability - Langfuse - 28k+ stars - Arize Phoenix - 14k+ stars - Helicone - 8k+ stars ### How to build a stack A minimal production stack in 2026: LangGraph (orchestration) + mem0 (memory) + Qdrant (retrieval) + Langfuse (observability) + one MCP server for your external systems. That covers planning, remembering, searching, debugging, and connecting. ### FAQ **What is the cheapest stack?** All of the above are open-source and self-hostable on a single 4-core server. **Do I need all six layers?** No. Start with framework + retrieval, add memory when your agent loses context, add observability before you deploy.