Agent-Reach
An open-source network-access capability layer for AI agents. Through a unified CLI and MCP protocol, it provides AI agents with capabilities including web scraping, YouTube subtitle retrieval, RSS feed reading, full-web search, GitHub repository access, and Bilibili content access. It is designed to work out of the box with zero API costs.
Core Facts
Timeline (last 90 days)
Agent-Reach provides six zero-configuration channels: any webpage, YouTube subtitles, RSS feeds, web-wide search, open-source repositories, and Bilibili.
Agent-Reach can be installed by sending a single natural-language instruction to an Agent, supporting agents such as Cloud Code, OpenCloud, and Cursor.
Agent-Reach uses the command agent-reach doctor for a health check, which lists the current backend used by each channel, which channels are working and which are not, and how to fix them.
The search capability in Agent-Reach is accessed via MCP, running as an MCP Server, and can perform web-wide retrieval without requiring an API Key.
yt-dlp has relatively limited support for Bilibili, so the Bilibili component of Agent-Reach is handled by a dedicated bili-cli, which can search and read video details without requiring login.
Twitter, Xiaohongshu, and Reddit require login authorization in Agent-Reach; Reddit's anonymous API has been blocked and there is no zero-configuration path.
Agent-Reach adopts a multi-backend routing design, with each platform configured with a 'primary + fallback' backend that automatically switches when one path fails.
What Agent-Reach provides for free is the scraping and reading channels, not bypassing each platform's official paid API; there is zero cost on a local computer, and only server deployment requires a proxy, at approximately $1 per month.
Agent-Reach's cookies are stored only in the local config.yaml with permissions set to 600, are not uploaded or transmitted externally, and the code is open source.
Using Agent-Reach scripts to call platforms carries a risk of detection and account suspension; the article recommends using a dedicated secondary account and controlling scraping frequency to avoid Bilibili's 412 blocking.
All Facts (11)
Obtaining social media data by bypassing official APIs may violate platform terms of service and carries legal risks in commercial scenarios.
65%UnverifiedAgent-Reach provides six zero-configuration channels: any webpage, YouTube subtitles, RSS feeds, web-wide search, open-source repositories, and Bilibili.
50%UnverifiedAgent-Reach can be installed by sending a single natural-language instruction to an Agent, supporting agents such as Cloud Code, OpenCloud, and Cursor.
50%UnverifiedAgent-Reach uses the command agent-reach doctor for a health check, which lists the current backend used by each channel, which channels are working and which are not, and how to fix them.
50%UnverifiedThe search capability in Agent-Reach is accessed via MCP, running as an MCP Server, and can perform web-wide retrieval without requiring an API Key.
50%Unverifiedyt-dlp has relatively limited support for Bilibili, so the Bilibili component of Agent-Reach is handled by a dedicated bili-cli, which can search and read video details without requiring login.
50%UnverifiedTwitter, Xiaohongshu, and Reddit require login authorization in Agent-Reach; Reddit's anonymous API has been blocked and there is no zero-configuration path.
50%UnverifiedAgent-Reach adopts a multi-backend routing design, with each platform configured with a 'primary + fallback' backend that automatically switches when one path fails.
50%UnverifiedWhat Agent-Reach provides for free is the scraping and reading channels, not bypassing each platform's official paid API; there is zero cost on a local computer, and only server deployment requires a proxy, at approximately $1 per month.
50%UnverifiedAgent-Reach's cookies are stored only in the local config.yaml with permissions set to 600, are not uploaded or transmitted externally, and the code is open source.
50%UnverifiedUsing Agent-Reach scripts to call platforms carries a risk of detection and account suspension; the article recommends using a dedicated secondary account and controlling scraping frequency to avoid Bilibili's 412 blocking.
50%