agent-reach

Research Anything Across 15 Platforms

Turn one research request into parallel pulls from 15 platforms including Twitter, Reddit, YouTube, GitHub, Bilibili, Xiaohongshu, and LinkedIn, each routed through whichever backend works right now.

Install the skill or give me an instruction how to install it - agent-reach from https://github.com/Panniantong/Agent-Reach

Free Open source Works in Claude, Cursor, Codex & more

Action Research

What it does

Ask one research question, or name a platform or paste a URL, in English or Chinese, and it pulls the actual content from up to 15 platforms at once: Twitter, Reddit, YouTube, GitHub, Bilibili, Xiaohongshu, LinkedIn, and more. Instead of a wall of HTML or a 403, you get the real material, posts and comment threads, article text, job listings, repo and PR data, video subtitles, and podcast transcripts, ready to synthesize.

The routing is what keeps it working. For each platform it picks whichever backend is live right now, an OpenCLI channel, a per-platform CLI, or a direct API, from a routing table, and follows built-in retry chains when one path fails rather than giving up. Six channels run with zero setup; the login-backed ones reuse your existing browser session. Run agent-reach doctor to see exactly what is healthy.

When to use it

  • You need raw content from many platforms at once

    One request pulls posts, threads, and transcripts from 15 platforms in parallel.

  • One platform's API keeps failing

    It routes each platform through whichever backend works right now and retries instead of guessing.

  • You're researching across English and Chinese sources

    It reaches Twitter, Reddit, and GitHub alongside Bilibili, Xiaohongshu, and more.

FAQ

What happens when a platform blocks access?

agent-reach checks each channel's health first and routes to whichever backend is working right now, following built-in retry chains when a command fails. It adapts instead of stopping at the first error.

Does it pull from platforms one at a time?

No. agent-reach collects from multiple platforms in parallel from a single request, so one question fans out across sources like Reddit, Twitter, and YouTube at once rather than in sequence.

How does it decide how to reach each platform?

agent-reach keeps a routing table and picks each platform's active backend, whether OpenCLI, a per-platform CLI, or a direct API. It chooses the path that works rather than guessing one.

Pairs well with