What Platforms Are Supported by Agent Reach? The Complete 2024 Guide

Agent Reach supports 15 distinct internet platforms—including YouTube, GitHub, Reddit, Twitter/X, and Bilibili—through a modular channel architecture that provides zero-configuration read access for public content while requiring authentication for private data and advanced social features.

Agent Reach is an open-source AI agent connectivity framework maintained in the Panniantong/Agent-Reach repository. The project implements a channel-based plugin system that abstracts platform-specific APIs into standardized read-and-search capabilities. Understanding what platforms are supported by Agent Reach requires examining the ALL_CHANNELS registry and the BaseChannel abstraction layer that powers the entire system.

Core Architecture: How Platform Support Works

The Channel Registry in agent_reach/channels/__init__.py

The single source of truth for supported platforms resides in agent_reach/channels/__init__.py. This file imports every channel class and aggregates them into the ALL_CHANNELS tuple (lines 9-42). The get_all_channels() function returns this complete list at runtime, enabling dynamic discovery of capabilities.

Each platform is implemented as a Python class inheriting from BaseChannel. When you execute an agent-reach command, the core router iterates through ALL_CHANNELS, invokes can_handle() on each instance, and delegates to the first matching channel's read() or search() method.

The BaseChannel Abstraction in agent_reach/channels/base.py

Every supported platform must implement the BaseChannel interface defined in agent_reach/channels/base.py. This contract requires four key methods:

• can_handle(url): Determines if this channel accepts the given URL or query
• read(url): Fetches and normalizes content from the platform
• search(query): Executes semantic or keyword search when available
• check(): Validates backend dependencies (e.g., yt-dlp, gh CLI, OpenCLI)

This architecture makes Agent Reach platforms plug-and-play. Adding support for a new service involves creating a subclass and registering it in __init__.py, without modifying core routing logic.

Zero-Configuration Platforms

These platforms require no authentication and work immediately after installing Agent Reach.

Web and RSS Feeds

• Web: The WebChannel provides read access to any public webpage using Jina Reader or direct fetching. No API keys or setup required.
• RSS: The RSSChannel (utilizing feedparser) parses any RSS or Atom feed automatically.

Video Platforms

• YouTube: The YouTubeChannel leverages yt-dlp to extract subtitles and metadata from public videos without API credentials.
• Bilibili: Partial support via BilibiliChannel allows searching and retrieving video details without configuration.

Developer Communities

• V2EX: Full read access to hot posts, node posts, detailed threads, and user profiles via V2EXChannel.
• GitHub (Public): The GitHubChannel reads public repository descriptions, files, and search indexes using the gh CLI without authentication.

Platforms Requiring Authentication

These platforms need additional configuration via OpenCLI, OAuth, or API keys to function.

Social Media Networks

• Twitter/X: Single tweet reading works immediately, but tweet search, timeline access, and long-thread reading require running "帮我配 Twitter" (configure Twitter for me).
• Reddit: No zero-config features available. You must install OpenCLI and run "帮我配 Reddit" to enable post reading, comment threads, and search.
• Facebook: Requires OpenCLI installation followed by "帮我配 Facebook" to access search, profiles, feeds, and groups.
• Instagram: Requires OpenCLI and "帮我配 Instagram" configuration for user search, profile viewing, and recent posts.
• Xiaohongshu: Requires OpenCLI and "帮我配 小红书" for search, content reading, and comment access.

Professional and Financial Platforms

• LinkedIn: Public page reading works via Jina Reader, but detailed profiles, company pages, and job search require "帮我配 LinkedIn" configuration.
• Xueqiu (Snowball): Chinese financial platform requiring "帮我配 雪球" for stock quotes, hot posts, and rankings.
• Xiaoyuzhou (小宇宙播客): Podcast platform requiring "帮我配 小宇宙播客" and utilizing Whisper for audio-to-text transcription.

Advanced GitHub Features

While public repos work immediately, accessing private repositories, Issues, PRs, and Forks requires OAuth authentication via "帮我登录 GitHub".

Semantic Search

Full-text search via ExaSearchChannel (using Exa AI) requires MCP (Model Context Protocol) auto-configuration, though this is handled automatically when the mcporter backend is available.

Diagnostics and Channel Verification

The agent_reach/doctor.py module provides runtime validation of all supported platforms. When you run diagnostics, the system probes each channel's check() method to verify backend dependencies are available.

# Verify which platforms are currently functional

agent-reach doctor

This command iterates over ALL_CHANNELS, tests each implementation, and reports which upstream tools (yt-dlp, OpenCLI, Exa) are detected and active.

Practical Usage Examples

Each platform maps to a specific channel class and backend tool:

# YouTube: Extract subtitles automatically (uses yt-dlp)

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

# GitHub: Read public repository metadata (uses gh CLI)

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

# RSS: Parse feed content (uses feedparser)

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

# Reddit: Requires configuration first

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

# Exa Search: Semantic web search (uses mcporter/Exa)

agent-reach search "latest LLM framework comparison"

Summary

• Agent Reach supports 15 platforms through the ALL_CHANNELS registry in agent_reach/channels/__init__.py
• Zero-config platforms include Web, YouTube, RSS, V2EX, and public GitHub repos
• Authenticated platforms require OpenCLI or OAuth setup for Reddit, Facebook, Instagram, Xiaohongshu, LinkedIn, Xueqiu, and Xiaoyuzhou
• Technical implementation relies on the BaseChannel interface with standardized can_handle, read, search, and check methods
• Runtime diagnostics via agent-reach doctor verify which channels are active based on available upstream tools

Frequently Asked Questions

How do I add support for a new platform in Agent Reach?

Create a new class inheriting from BaseChannel in agent_reach/channels/your_platform.py, implementing the required can_handle, read, search, and check methods. Import this class in agent_reach/channels/__init__.py and append it to the ALL_CHANNELS tuple. The agent-reach doctor command will automatically detect and test your new channel.

Why does Reddit require OpenCLI while YouTube works immediately?

YouTube support relies on yt-dlp, a standalone tool that extracts public video metadata without API authentication. Reddit, however, requires authenticated API access for even basic reading due to platform restrictions. The RedditChannel delegates to OpenCLI (or rdt-cli) to handle OAuth tokens securely, whereas YouTubeChannel uses direct HTTP extraction with yt-dlp.

How can I check if a specific platform channel is working?

Run agent-reach doctor to execute the diagnostic engine. This command calls the check() method on every channel in ALL_CHANNELS, verifying that required binaries (like gh, yt-dlp, or OpenCLI) are installed and accessible. The output indicates which platforms are ready for use and which require configuration.

Does Agent Reach support writing data to platforms or only reading?

Currently, Agent Reach platforms are read-only and search-capable. The BaseChannel interface defines read() and search() methods, but no write() or post() abstraction exists in the current implementation. All supported channels—GitHub, Twitter/X, Reddit, and others—focus on ingesting content into AI agent contexts rather than publishing to those platforms.

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