A high-performance proxy skill that converts web pages and search results into LLM-friendly Markdown to optimize token consumption.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install deshell
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 deshell using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
DeShell is a powerful utility designed to streamline how AI agents interact with the web by converting complex web pages into clean, LLM-friendly Markdown. As one of the most efficient Openclaw Skills for data retrieval, it helps developers save significantly on token costs—often reducing consumption by 60–80% compared to raw HTML extraction.
By serving as a sophisticated proxy, DeShell eliminates the need for manual URL construction or complex header management. It provides a consistent interface for searching the web and fetching content, ensuring that your agents receive high-quality, structured information every time they interact with online resources.
To integrate this tool into your collection of Openclaw Skills, follow these installation steps:
npm install @deshell/mcp
export DESHELL_API_KEY='your_api_key_here'
The skill organizes data retrieval through a simple CLI interface and returns structured text.
| Attribute | Detail |
|---|---|
| Primary Output | Cleaned Markdown (UTF-8) |
| Search Metadata | Includes Title, URL, and Description per result |
| Configuration | Managed via DESHELL_API_KEY and DESHELL_PROXY_URL |
| Extensibility | Custom headers supported via DESHELL_EXTRA_HEADERS |
Loading
A mandatory security wrapper for strict-mode execution, evidence logging, and policy enforcement in autonomous operations.

A zero-dependency, GPU-native programming language that converts natural language descriptions into high-performance Vulkan compute programs.

A sophisticated AI mentor that uses longitudinal memory and philosophical frameworks to guide personal development and maintain compassionate accountability.

A specialized project management agent that protects hyperfocus and minimizes context-switching costs for ADHD-driven engineering workflows.

Distil enables AI agents to search the web and fetch web pages as clean, LLM-friendly Markdown to drastically reduce token consumption.

A comprehensive API skill for scheduling and managing social media content across 13 major platforms from a single interface.








































