screenpipe/screenpipe

Screenpipe is a local screen and audio recording tool that captures your computer activity continuously and provides context to AI agents and applications.

The tool solves the problem of giving AI agents meaningful context about what you are actually doing on your computer. It records screen and audio locally on your machine, then makes this captured history searchable and queryable through natural language. Agents like Claude can ask questions about your recent activity, summarize conversations, or generate automations based on patterns in your work. The approach keeps all data local by design, storing recordings and indexed content only on your device rather than sending it to external services.

Screenpipe suits developers and teams who want to build agent-driven automations grounded in real computer activity, or who need to search through their work history using natural language queries. It works as both a desktop application with auto-updates and as a CLI tool that coding agents can integrate with directly. The tool is designed for scenarios where you want AI to understand context from your actual workflow—whether that is generating task automations, creating agent skills, or answering questions about what you did in the last few hours. The README emphasizes that this is source-available software you can inspect and modify, and positions it as a way for teams to share knowledge without creating surveillance infrastructure.

Maintainers respond to issues and pull requests within a few days. Work in the issue tracker centers on bug fixes, enhancements, and medium-priority items.