CamelAGI

CamelAGI provides a versatile platform for managing personal computation agents, offering control from Telegram, web applications, or your local terminal. Key features include: * Self-hosted multi-agent management * Secure control of local computation environments via remote commands * Integrated two-tier memory for agents * Support for external tool servers (MCP) * Admin and per-agent Telegram bots for seamless interaction Built upon the Claude Agent SDK, CamelAGI enhances your interaction with local computation tools. It allows you to send commands to your locally installed computation environments, ensuring that authentication and sensitive operations remain on your device, leveraging your existing Anthropic subscription. The platform emphasizes a lightweight codebase, making it easier to maintain and adapt. Agents within CamelAGI benefit from isolated two-tier memory, combining curated `MEMORY.md` files with daily auto-journaling and recency-boosted search. This ensures agents retain context and learn effectively over time. The system also supports voice transcription for convenient input via voice messages, processed into text using services like Groq, OpenAI, or Deepgram. CamelAGI is ideal for developers, researchers, and power users who require secure, self-hosted control over their computation resources. It's particularly useful for those who need to manage multiple specialized agents and integrate external tools, all accessible from mobile devices or a desktop terminal.