OpenBiliClaw is a local-first open-source AI content discovery agent that proactively finds content across multiple platforms based on user preferences.
The tool addresses content discovery by learning user interests and then autonomously searching across platforms including Bilibili, Xiaohongshu, Douyin, YouTube, X, Zhihu, Reddit, Weibo, and the open web. It operates as a self-evolving agent that prioritizes privacy by running locally rather than relying on cloud services. The system can be extended through deepseek harness plugins, allowing customization of its discovery behavior and integration with other tools.
Developers should choose this tool if they want a privacy-preserving alternative to centralized recommendation systems and need content discovery across Chinese and international platforms. It suits projects where users want an AI agent that understands their preferences over time and actively surfaces relevant content rather than waiting for manual searches. The local-first architecture means no user data leaves the user's machine, making it appropriate for privacy-sensitive deployments. The project includes both a Python backend and TypeScript components, including a Chrome extension for browser integration.
The project shows active development with ongoing refinements to its core discovery mechanisms and platform integrations. Work continues on expanding the range of supported platforms and improving the agent's ability to understand and predict user interests. The codebase demonstrates attention to making the system extensible through its plugin architecture, allowing community contributions to add new platforms or discovery strategies.
