#777 · Primary category: AI Agents & Automation

phantom

ai-agents ai-coworker anthropic autonomous-agents bun claude docker llm mcp self-evolution slack typescript

An AI co-worker with its own computer. Self-evolving, persistent memory, MCP server, secure credential collection, email identity. Built on the Claude Agent SDK.

Project last updated:06/16/26

GitHub Stars

1.5K

Forks

194

Contributors

7

License

Apache-2.0

Why we included this project

Most agent projects hand you a chat window and lose the context when you close it. Phantom runs as a persistent service on its own machine, so it can install software, stand up databases, and leave dashboards running after the conversation ends, all while keeping long-term memory and an MCP server for exposing tools. It can take on a Slack identity or its own email address, and the underlying model is swappable across Anthropic, OpenRouter, Ollama, vLLM, and others, so you can change brains without losing memory or tools. The self-evolution loop is the part worth attention: one production instance loaded 28.7 million rows into ClickHouse, assembled a full analytics stack, and then began monitoring its own infrastructure without being told to. Teams running long-lived autonomous agents, or individuals who want a co-worker they can point at recurring tasks, will find a genuinely different deployment model here.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category