Agent Reach

// Overview

Agent Reach is an MIT-licensed capability layer that gives a CLI AI agent access to the open internet. Rather than being one more scraper, it is the selection-and-plumbing layer above them: for each platform it picks a working method, installs it, health-checks it, and routes requests to a primary backend with fallbacks behind it — so when a site changes or an endpoint is deprecated, the agent keeps working.

Coverage splits into two tiers. Web pages, YouTube, RSS, GitHub, Bilibili and semantic web search work with zero configuration and no API keys. Twitter/X, Reddit, Facebook, Instagram and LinkedIn need you to be logged in, and Xiaohongshu, V2EX and Xueqiu sit in between. Credentials are kept locally in `~/.agent-reach/config.yaml` and never leave the machine, and the underlying tools it installs are themselves open source, so there are no per-call API fees.

Installation is deliberately agent-driven: you hand your agent a link to the project's install document and it does the rest, which is why the project describes itself as agent-agnostic — it works with Claude Code, OpenClaw, Cursor, Windsurf or any CLI-capable agent. A built-in `agent-reach doctor` command reports which channels currently work and why the others don't, and Exa semantic search is wired in through mcporter as a free, key-less MCP backend.

// What it does

• One capability layer that selects, installs and health-checks a working reader for each platform
• Zero-config access to web pages, YouTube, RSS, GitHub, Bilibili and semantic web search
• Login-based access to Twitter/X, Reddit, Facebook, Instagram, LinkedIn, Xiaohongshu, V2EX and Xueqiu
• Multi-backend routing — every platform has a primary plus fallbacks, so a site change doesn't break the agent
• `agent-reach doctor` self-diagnostic that reports which channels work and why the rest don't
• Credentials stay local in `~/.agent-reach/config.yaml`; the underlying tools are open source, so there are no API fees
• Agent-agnostic — works with Claude Code, OpenClaw, Cursor, Windsurf or any CLI-capable agent

// Getting started

Agent Reach is installed by the agent itself: you paste one instruction pointing at the project's install document, and the agent performs the setup.

Install Agent Reach for me: https://raw.githubusercontent.com/Panniantong/agent-reach/main/docs/install.md

agent-reach doctor

agent-reach install --system

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Give a coding or research agent read access to social platforms it otherwise cannot open
• Pull YouTube transcripts, RSS feeds and GitHub content into an agent workflow without writing a scraper
• Run web research from an agent without paying per-call search-API fees
• Keep an agent's web access working when a platform changes, via the fallback backends and doctor check

// How Agent Reach compares

Agent Reach alongside other open-source web scraping & crawling tools AI/TLDR tracks, ranked by GitHub stars.