Agent Reach

Agent Reach is an open-source CLI with 100,000+ stars on GitHub that gives your AI agent eyes on the entire internet — reading and searching social platforms and the web with zero API fees.

Overview

Agents are only as useful as what they can see, and most of the interesting internet sits behind platforms that demand paid API access. Agent Reach takes a different route: one Python CLI that lets your agent read and search Twitter/X, Reddit, YouTube, GitHub, Bilibili and Xiaohongshu, plus general web pages, scraped through accessible channels rather than metered official APIs. It is built as agent infrastructure — a companion skill for tools like Claude Code and Cursor, and an MCP-friendly building block — so an agent can pull a video transcript, scan a subreddit’s sentiment or research a topic across platforms in a single workflow. The project is MIT licensed and moving fast, with repository updates in October 2026 and a star count past 100,000 driven by the agent-building community.

Key Features

• Multi-platform reading – Twitter/X, Reddit, YouTube, GitHub, Bilibili and Xiaohongshu from one CLI.
• Zero API fees – avoids paid platform APIs by using accessible scraping channels.
• Transcripts and search – pull YouTube transcripts and search across sources programmatically.
• Agent-native design – ships as a skill for Claude Code and Cursor, with MCP integration paths.
• Simple Python CLI – install, configure once, and call it from any agent workflow.
• MIT licensed – free for personal and commercial agent projects.

💳 Pricing

Agent Reach is completely free and open source under the MIT license — its whole premise is research access without API fees. Your agent’s own model costs still apply.

Who Is It For

Agent builders and power users who want their AI assistants to research the live social web — tracking discussions, gathering sources and monitoring platforms — without paying for a stack of platform APIs.