Agent Reach provides AI agents with high-fidelity eyes across the internet, enabling seamless searching and reading of over 14 major social and content platforms.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install modesty-agent-reach
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
Copy this prompt to OpenClaw to install it automatically.
Help me install modesty-agent-reach using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
Agent Reach is a robust connectivity layer designed to grant AI agents comprehensive access to the modern web. By integrating with Openclaw Skills, it bridges the gap between LLMs and real-time data from platforms like Twitter/X, Reddit, YouTube, and GitHub. It utilizes specialized upstream tools and the SkillBoss API Hub to bypass common bot-detection hurdles, ensuring your agent can retrieve fresh, relevant information from across the global internet.
Beyond simple web scraping, Agent Reach understands the nuances of different platforms, offering specialized handlers for Chinese social media like WeChat, Bilibili, and XiaoHongShu. It is an essential tool for developers building research-heavy or social-aware AI agents that require more than just static training data or basic search engine results.
To install the core skill and check dependencies, use the following commands:
# Install the agent-reach utility
npm install -g agent-reach
# Run the diagnostic tool to check channel availability
agent-reach doctor
# Configure your API Key for web scraping and search
export SKILLBOSS_API_KEY="your_key_here"
For specific channels like Twitter, ensure undici is installed via npm install -g undici.
Agent Reach follows a strict workspace rule to maintain cleanliness. Data is organized as follows:
| Type | Path | Description |
|---|---|---|
| Temporary | /tmp/ |
Used for transient outputs like video subtitles (.vtt) or temporary scrapes. |
| Persistent | ~/.agent-reach/ |
Stores configuration, tool-specific data, and persistent environment settings. |
| Metadata | JSON | Most CLI tools support a --json flag to return structured metadata for agent parsing. |
All scraped web content is typically returned as a Markdown string to optimize LLM context window usage.
agent-reach doctor for real-time troubleshooting.agent-reach configure proxy.Loading
A professional command-line interface for accessing real-time and historical e-commerce data across Amazon, Shopee, and Walmart.

A production-ready framework for managing multi-agent teams with defined roles, structured task lifecycles, and rigorous handoff protocols.

A professional-grade intelligent trading assistant dedicated to providing real-time data and strategic analysis for the Chinese A-share market.

A utility for removing stale or ghost subagent sessions from the OpenClaw webchat sidebar to maintain a clean workspace.

Wiggle Rooms is a filesystem-mediated chat protocol that allows AI agents to communicate in shared rooms by simply reading and writing to a markdown file.

A powerful AI agent skill for interacting with decentralized SOLID Pods to manage RDF data, containers, and granular access controls.








































