Give AI coding assistants the ability to research topics on social media and summarize findings as part of a workflow.
Extract video subtitles and metadata from YouTube without paying for API access.
Pull live data from Twitter, Reddit, and GitHub to feed into automated analysis or reporting tools.
Build web-scraping workflows that respect platform terms by using local browser authentication instead of server-side API calls.
Requires browser cookies to be extracted and configured for each platform you want to use.
Agent Reach is a Python command-line toolkit that gives AI agents the ability to read and search the internet, specifically Twitter/X, Reddit, YouTube, GitHub, Bilibili, and XiaoHongShu, without paying for expensive APIs. The problem it solves is that AI coding assistants like Claude Code or Cursor can write code and manage files, but they go blind when asked to browse social media, extract video subtitles, or search the web, because those platforms either charge for API access, block server requests, or require logins. Instead of building direct integrations, Agent Reach works as a scaffolding layer: it installs and wires together a set of best-in-class open-source command-line tools such as yt-dlp for YouTube subtitles, twitter-cli for Twitter, rdt-cli for Reddit, gh CLI for GitHub, and feedparser for RSS. Once installed, an AI agent can simply call these tools from the terminal. Platforms that require login use browser cookies, which stay stored locally on your own machine and are never uploaded anywhere. A built-in "doctor" command checks which platforms are working and which need configuration. Each platform's integration is a swappable module, so you can replace any underlying tool with an alternative you prefer. You would use Agent Reach when you want your AI agent to research topics on social platforms, summarize videos, or pull live data from the web as part of a workflow, all without paying API fees.
Generated 2026-05-18 · Model: sonnet-4-6 · Verify against the repo before relying on details.