Menu
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
npx skills add https://github.com/javimosch/open-claw-skills --skill deshellThe command stays on one line. Scroll horizontally to inspect it before copying.
Prefer a local copy? Download the files currently available to SkillsMP.
Based on SOC occupation classification
Showing SKILL.md
| name | deshell |
| description | Fetch web pages as clean Markdown and search the web via the DeShell proxy |
| version | 1.5.0 |
| metadata | {"openclaw":{"emoji":"๐","requires":{"bins":"[Truncated]","env":"[Truncated]"},"primaryEnv":"DESHELL_API_KEY","install":["[Truncated]"]}} |
Gives agents discoverable, consistent access to the DeShell proxy โ no manual URL construction, no remembering headers or API keys.
DeShell converts web pages into clean Markdown, saving 60โ80% of tokens for LLM consumption.
Get your API key from https://deshell.ai (sign up or use your existing key)
Install the deshell CLI manually (one-time):
npm install @deshell/mcp
Set the DESHELL_API_KEY environment variable
# Fetch any URL as clean Markdown
deshell fetch https://example.com
# Search the web and get results as Markdown
deshell search "best practices for Go error handling"
# Multi-word queries work naturally โ no quoting needed
deshell search top 10 AI companies 2025
# Take a screenshot of a web page and return it as an image
deshell screenshot https://example.com
# Render a web page (such as a single page javascript app) before trying to extract markdown
deshell render https://example.com
# Fetch a URL and return its raw content bypassing any attempt to render markdown
deshell raw https://example.com
# Fetch a URL and return its content without using the cache
deshell nocache https://example.com
| Environment Variable | Default | Description |
|---|---|---|
DESHELL_API_KEY | (none) | API key |
DESHELL_PROXY_URL | https://proxy.deshell.ai/ | Proxy base URL |
DESHELL_EXTRA_HEADERS | (none) | Comma-separated extra headers in Header-Name:value format |
deshell fetch โ returns page content as Markdown on stdoutdeshell search โ returns search results with titles, URLs, descriptions, and page content as Markdown on stdoutTo pass additional DeShell headers, use comma-separated Header-Name:value pairs:
# Single header
DESHELL_EXTRA_HEADERS="X-DeShell-No-Cache:true" deshell fetch https://example.com
# Multiple headers
DESHELL_EXTRA_HEADERS="X-DeShell-No-Cache:true,X-DeShell-Max-Tokens:2000" deshell fetch https://example.com
# Research a topic
deshell search "OpenClaw agent framework"
# Read documentation
deshell fetch https://docs.github.com/en/rest
# Force fresh fetch (bypass cache)
DESHELL_EXTRA_HEADERS="X-DeShell-No-Cache:true" deshell fetch https://news.ycombinator.com
If you prefer not to install npm packages, you can call the proxy directly with curl:
# Fetch a page
curl -s "https://proxy.deshell.ai/https://example.com" \
-H "X-DeShell-Key: YOUR_API_KEY"
# Search the web
curl -s "https://proxy.deshell.ai/search?q=your+query" \
-H "X-DeShell-Key: YOUR_API_KEY" \
-H "Accept: text/markdown"