Agent Reach Supported Platforms: YouTube, GitHub, Twitter, and 12 More Integrations

Agent Reach supports 15 distinct platforms including YouTube, GitHub, Twitter/X, Reddit, LinkedIn, Bilibili, RSS feeds, and semantic web search, offering zero-configuration read access for public content and authenticated capabilities for private data retrieval.

Agent Reach is an open-source Python framework that extends AI agent capabilities with read-and-search access across internet services. The platform registry defined in agent_reach/channels/__init__.py maintains the canonical list of supported integrations through the ALL_CHANNELS registry. Understanding the supported platforms helps developers configure the correct authentication methods and leverage zero-setup features for immediate deployment.

Platform Registry Architecture

The core of Agent Reach’s multi-platform support lives in agent_reach/channels/__init__.py, where the ALL_CHANNELS list (lines 9-42) imports and registers every available channel class. The function get_all_channels() returns this complete registry at runtime, enabling dynamic discovery of capabilities.

Each platform is implemented as a concrete class inheriting from BaseChannel in agent_reach/channels/base.py. This abstract base class enforces a consistent contract requiring four methods: can_handle() to test URL compatibility, read() to fetch content, search() to query the platform, and check() to verify backend availability. When an agent requests a capability, the core router selects the appropriate channel and forwards requests to underlying tools like yt-dlp, OpenCLI, or gh CLI without additional wrapping.

The agent-reach doctor command (implemented in agent_reach/doctor.py) iterates over ALL_CHANNELS, probes each backend in order, and reports which implementations are currently functional, making it easy to diagnose configuration issues across the supported platforms.

Zero-Configuration Platforms

Several Agent Reach supported platforms require no authentication or setup, providing immediate read access to public data:

• Web – Reads any public webpage via Jina Reader integration.
• YouTube – Extracts subtitles and metadata using yt-dlp (implemented in agent_reach/channels/youtube.py).
• RSS – Parses any RSS or Atom feed using feedparser (implemented in agent_reach/channels/rss.py).
• V2EX – Retrieves hot posts, node discussions, detailed replies, and user profiles without API keys.
• Xueqiu (Snowball) – Accesses stock quotes, search results, hot posts, and market rankings.

Platforms Requiring Authentication

Advanced features and private data access require configuration through OpenCLI or platform-specific credentials:

• GitHub – Public repository reading works immediately; private repos, Issues, PRs, and Forks require authentication via the gh CLI (implemented in agent_reach/channels/github.py). Trigger configuration with the command “帮我登录 GitHub”.
• Twitter/X – Single tweet reading requires no setup, but tweet search, timeline access, and long-thread reading need OpenCLI configuration (“帮我配 Twitter”).
• Bilibili – Video search and details work out-of-the-box; subtitles require OpenCLI setup (“帮我配 B站”).
• Reddit – Post and comment reading, plus search capabilities, require OpenCLI installation and configuration (“帮我配 Reddit”) as implemented in agent_reach/channels/reddit.py.
• Facebook – Search, profile access, feeds, and groups require OpenCLI (“帮我配 Facebook”).
• Instagram – User search, profile viewing, recent posts, and Explore access need OpenCLI (“帮我配 Instagram”).
• Xiaohongshu – Search, content reading, and comment access require OpenCLI (“帮我配 小红书”).
• LinkedIn – Public pages are readable via Jina Reader; profile details, company pages, and job search require configuration (“帮我配 LinkedIn”).
• Xiaoyuzhou (小宇宙播客) – Podcast audio-to-text transcription requires Whisper setup (“帮我配 小宇宙播客”).

Semantic Search Capabilities

Agent Reach includes full-text semantic search via the ExaSearchChannel class in agent_reach/channels/exa_search.py. This capability is auto-configured via MCP (Exa) and enables AI agents to perform intelligent web searches beyond simple keyword matching, retrieving relevant content without platform-specific authentication.

CLI Diagnostics and Usage Examples

Verify which Agent Reach supported platforms are active and interact with them using these commands:

# Diagnose all channels – shows which back-ends are functional

agent-reach doctor

# Read a public YouTube video (auto-extract subtitles)

agent-reach read https://www.youtube.com/watch?v=dQw4w9WgXcQ

# Search the web via Exa (semantic search)

agent-reach search "latest LLM framework comparison"

# Fetch a GitHub repository description

agent-reach read https://github.com/Panniantong/Agent-Reach

# Grab the latest posts from an RSS feed

agent-reach read https://news.ycombinator.com/rss

# Configure Reddit access (requires OpenCLI install), then read a thread

agent-reach configure reddit
agent-reach read https://www.reddit.com/r/python/comments/xyz123/example_thread/

Each command maps to a specific channel implementation: YouTubeChannel uses yt-dlp, GitHubChannel wraps the gh CLI, RSSChannel utilizes feedparser, and RedditChannel interfaces with rdt-cli or OpenCLI.

Summary

• Agent Reach supports 15 distinct platforms ranging from social media (Twitter, Instagram, Facebook) to developer tools (GitHub) and Chinese services (Bilibili, Xueqiu, Xiaohongshu).
• The platform registry in agent_reach/channels/__init__.py maintains the ALL_CHANNELS list, enabling dynamic discovery via get_all_channels().
• Five platforms (Web, YouTube, RSS, V2EX, Xueqiu) work with zero configuration using backends like yt-dlp and feedparser.
• Nine platforms require authentication through OpenCLI or specific tools (gh CLI, Whisper) to access private data or advanced search features.
• The agent-reach doctor command probes all channels to report real-time availability of backend dependencies.

Frequently Asked Questions

How do I check which Agent Reach platforms are currently active?

Run the agent-reach doctor command from the terminal. This diagnostic tool iterates through the ALL_CHANNELS registry defined in agent_reach/channels/__init__.py, calls the check() method on each BaseChannel subclass, and reports which backends (such as yt-dlp, gh, or OpenCLI) are installed and functional.

Does Agent Reach require API keys for all platforms?

No. Five platforms work without any configuration: Web (via Jina Reader), YouTube (via yt-dlp), RSS feeds (via feedparser), V2EX, and Xueqiu. However, platforms like GitHub, Twitter/X, Reddit, and LinkedIn require authentication tokens or OpenCLI installation to access private data, search functions, or timeline features.

Can Agent Reach access private GitHub repositories?

Yes, but authentication is required. The GitHubChannel class in agent_reach/channels/github.py supports public repository reading immediately, but accessing private repos, Issues, and PRs requires configuring the gh CLI tool. Use the command “帮我登录 GitHub” or agent-reach configure github to initiate the authentication flow.

What backend tools does Agent Reach use for YouTube and Reddit?

Agent Reach uses yt-dlp for YouTube operations (subtitle extraction and metadata reading) as implemented in agent_reach/channels/youtube.py. For Reddit, the RedditChannel class in agent_reach/channels/reddit.py interfaces with OpenCLI or the rdt-cli tool to fetch posts, comments, and perform searches.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}