Skip to content
HUIAHUB

Panniantong/Agent-Reach

Agent-Reach

Give your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.

Metrics updated Established+7.7% 7d
View on GitHub (opens in a new tab)

GitHub metrics

Stars
97K
+6.9K this week
Forks
8.5K
30-day growth
—
Open issues
92
90K93K97KOct 4, 2026Oct 13, 2026
GitHub stars from the official GitHub API, recorded each time the repository was checked (3 days).
GitHub stars by day
DayGitHub stars
2026-10-0489795
2026-10-1095251
2026-10-1396741

What it does

Agent Reach is a Python command-line tool that prepares an AI agent to read and search sites that are awkward to reach from a script, including Twitter/X, Reddit, YouTube, Bilibili, XiaoHongShu and GitHub. It leaves the actual reading to other tools. For each platform it picks an upstream project (Jina Reader for web pages, yt-dlp for YouTube, gh for GitHub, feedparser for RSS, twitter-cli and OpenCLI for social sites), installs it, checks that it works and gives the agent a SKILL.md file describing what to call.

Every platform has an ordered list of backends, and when one breaks the maintainers move the next one up. The README cites June 2026, when Bilibili blocked yt-dlp and the tool switched to bili-cli. Running agent-reach doctor shows which channels work and which backend each is using.

Web pages, YouTube, RSS, public GitHub repositories, V2EX and basic Bilibili search need no setup. Twitter, Reddit, XiaoHongShu, Facebook and Instagram need a logged-in session, either from cookies you export by hand or through OpenCLI reusing your desktop Chrome login. The main README is in Chinese; English, Japanese and Korean versions sit in the docs folder.

Why it matters

Each platform puts up its own barrier for agents: paid APIs, blocked server IPs, mandatory logins or HTML that is hard to read. Agent Reach bundles a working route for each into one installer with a health check, and the maintainers say they will keep those routes working as platforms change. The README says every tool and API it relies on is free, and the only possible cost is a proxy of about $1 a month on servers or restricted networks.

Main features

  • Reads any web page as clean Markdown through Jina Reader
  • YouTube subtitles and search through yt-dlp
  • Bilibili search and video details without login via bili-cli
  • Twitter/X, Reddit, Facebook and Instagram through cookies or OpenCLI browser sessions
  • Web search through Exa, connected with mcporter
  • agent-reach doctor reports each channel's status and active backend
  • Install only inspects the machine unless you pass `--system`; `--dry-run` previews changes
  • Uninstall command that also removes stored tokens and cookies

Who it's for

  • Users of Claude Code, Cursor, OpenClaw or any agent that runs shell commands and needs web or social content
  • Researchers following discussion on Twitter, Reddit, YouTube or Chinese platforms such as Bilibili, XiaoHongShu and V2EX
  • Developers who would rather use free open-source scrapers than paid platform APIs

Strengths

  • Cookies and tokens stay in ~/.agent-reach/config.yaml with owner-only (600) file permissions, per the README
  • Cautious default: install changes nothing on the system unless `--system` is passed
  • Backends can be swapped per channel, so one broken scraper leaves the others working
  • MIT licensed and built on open-source upstream tools

Limitations

  • Cookie-based access can get accounts restricted or banned; the README advises using a dedicated secondary account
  • The recommended install asks your agent to fetch and follow the instructions in a Markdown file from the repository
  • Reddit has no zero-config route, and Facebook and Instagram work only on desktop through OpenCLI
  • The README warns that a package with the same name on PyPI is a different project

Getting started

Requirements: Python 3.10 or later and an agent that can run shell commands; some channels use Node.js, the gh CLI and mcporter, and logged-in accounts are needed for Twitter, Reddit, XiaoHongShu, Facebook and Instagram.

Always follow the official instructions in the README on GitHub. This overview is our summary, not a copy.

Sources