A sophisticated orchestration layer that transforms raw web fetch results into secure, policy-compliant, and normalized Markdown for AI agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install markdown-browser
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install markdown-browser using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Markdown Browser?
The Markdown Browser skill serves as a specialized post-processing wrapper designed to work in tandem with official web fetching tools within the Openclaw Skills ecosystem. Rather than replacing the fetch layer, it introduces a MECE (Mutually Exclusive, Collectively Exhaustive) architecture to handle the complex transformation of web data into agent-ready context. It focuses on maintaining the technical truth of the source while applying rigorous policy and privacy standards.
By acting as a bridge between raw network responses and downstream logic, this skill ensures that AI agents receive clean, high-quality data. It manages everything from HTML-to-Markdown conversion to computing policy actions based on content signals, making it an indispensable tool for developers building resilient Openclaw Skills that require reliable web data processing.
Markdown Browser Use Cases
- Normalizing diverse web content types into consistent Markdown for LLM consumption.
- Implementing privacy-aware data pipelines by redacting sensitive URL fragments and query parameters.
- Enforcing automated content policies like allow_input or block_input based on Content-Signal headers.
- Providing stable, structured output schemas for AI agents to prevent breaking changes in web data handling.
How Markdown Browser Works
- The process begins by calling the official web_fetch tool to retrieve raw page data.
- The resulting JSON payload, along with optional headers like Content-Signal, is passed into the Markdown Browser wrapper.
- The policy layer analyzes the headers to compute the appropriate policy_action (e.g., allow_input).
- The privacy layer redacts identifying information from the source URL while maintaining its structural utility for debugging.
- The normalization layer detects the content type and converts HTML to Markdown via Turndown if necessary.
- A structured object containing the normalized content, token estimates, and metadata is returned for agent use.
Markdown Browser Setup
To integrate this capability into your Openclaw Skills environment, navigate to the skill directory and install the production dependencies:
npm install --omit=dev
You can verify the functionality via the CLI by processing a pre-fetched JSON result:
node browser.js \
--input /tmp/web_fetch.json \
--content-signal "ai-input=yes, search=yes, ai-train=no" \
--markdown-tokens "1820"
Markdown Browser Data Schema & Taxonomy
The skill organizes its output into a stable schema to ensure compatibility across different Openclaw Skills. The primary tool, process_web_fetch_result, returns the following structure:
| Property | Type | Description |
|---|---|---|
content |
string | The final processed Markdown or text content. |
format |
string | The output format: markdown, html-fallback, or text. |
token_estimate |
number | The calculated token count for the content payload. |
policy_action |
string | The computed action: allow_input, block_input, or needs_review. |
source_url |
string | The redacted version of the original URL. |
fallback_used |
boolean | Indicates if the HTML-to-Markdown conversion was triggered. |
Markdown Browser Advanced Features
- MECE architectural design ensures no overlap between fetch, policy, privacy, and normalization logic.
- Sophisticated URL redaction that protects user privacy without losing the context of the data source.
- Automated token estimation to help agents manage context window constraints effectively.
- Dynamic policy computation that translates complex Content-Signal headers into simple actionable states for Openclaw Skills agents.
- Turndown-powered HTML normalization providing high-fidelity Markdown fallbacks for non-native Markdown sources.
SKILL.md
Loading
Related Openclaw Skills
Kimi Agent Policy
A standardized tool-use and behavior policy for OpenClaw agents based on Moonshot AI's Kimi orchestration rules.
HTTP Retry Skill
A robust networking utility that implements exponential backoff and timeout controls to prevent agent failures due to transient network errors.
EvoMap Node Controller
A comprehensive management tool for starting, stopping, and monitoring EvoMap evolver nodes across distributed server environments.
Cloudflare MCP
A powerful Model Context Protocol (MCP) integration that enables AI agents to manage over 2500 Cloudflare API endpoints including Workers, DNS, and storage.
Intelligence Ingestion
A strategic pipeline that automatically captures, analyzes, and stores external information into your local knowledge base and system memory.
OpenClaw Switch
A comprehensive model manager for handling multi-provider switching, automatic failover chains, and subagent routing.
Featured*
