A protocol-driven approach for AI agents to discover and execute structured skills directly from websites instead of relying on fragile web scraping.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install web-skills-protocol
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 web-skills-protocol 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 Web Skills Protocol (WSP)?
The Web Skills Protocol (WSP) provides a standardized way for AI agents to interact with web services by looking for machine-readable capability files. Instead of guessing UI elements or reverse-engineering complex JavaScript, an agent using Openclaw Skills can fetch a skills.txt file to understand exactly what actions a site supports.
This method ensures higher reliability, follows the site owner's intended usage, and provides structured API access for tasks like searching, ordering, or data retrieval. By prioritizing these defined protocols over traditional scraping, developers can build more resilient agents that respect site-specific rules and authentication requirements.
Web Skills Protocol (WSP) Use Cases
- Automated product searching on e-commerce sites without fragile DOM scraping.
- Executing complex transactions or orders through documented agent workflows.
- Discovering API endpoints and authentication requirements for web services dynamically.
- Reducing agent errors on high-interactivity or heavy JavaScript websites by using structured Openclaw Skills.
How Web Skills Protocol (WSP) Works
- Check the website's root origin for a skills.txt or agents.txt discovery file.
- Parse the discovery file to extract site descriptions, general notes, and specific skill entries.
- Match the user's specific request intent against the descriptions of the discovered skills.
- Fetch the specific SKILL.md file associated with the chosen functionality from the site's skills directory.
- Process the YAML frontmatter to identify base URLs, authentication types, and rate limits.
- Execute the instructions within the Markdown body to perform the requested action via structured API calls using the provided Openclaw Skills framework.
Web Skills Protocol (WSP) Setup
To enable Web Skills Protocol discovery in your agent environment, ensure your agent is configured to check for standardized discovery files before initiating a scrape.
# Example of checking for a discovery file
curl -I https://example.com/skills.txt
# If a 200 OK is returned, parse the markdown-based discovery file
# Fetch the corresponding SKILL.md for the desired task
curl https://example.com/skills/search/SKILL.md
Web Skills Protocol (WSP) Data Schema & Taxonomy
The protocol relies on a structured YAML frontmatter and Markdown body to define its capabilities:
| Attribute | Description | Required |
|---|---|---|
| name | Unique identifier for the skill | Yes |
| auth | Authentication method (none, api-key, bearer, oauth2) | Yes |
| base_url | The root endpoint for all skill-related API calls | Yes |
| rate_limit | Object containing agent (recommended) and api (hard) limits | No |
| instructions | Markdown body containing endpoints, parameters, and examples | Yes |
Web Skills Protocol (WSP) Advanced Features
- Automatic fallback to agents.txt if skills.txt is unavailable to maximize discovery of Openclaw Skills.
- Session-based caching of discovery files to minimize redundant network requests and improve latency.
- Granular rate limiting support that distinguishes between recommended agent pacing and hard API limits.
- Support for multiple authentication schemes including OAuth2 and Bearer tokens for secure site interaction.
- Intent-based skill matching for handling ambiguous user requests by comparing descriptions.
SKILL.md
Loading
Related Openclaw Skills
Git Commit Helper
An AI-powered utility for generating standardized git commit messages following the Conventional Commits specification.
ClankerKit
A comprehensive suite for AI agents to manage autonomous smart contract wallets, execute trades, and enforce spending policies on the Monad blockchain.
ClawdNet Agent Skill
Register, discover, and interact with decentralized AI agents through the ClawdNet registry.
Shadow Number
A powerful utility for AI agents to acquire disposable phone numbers and automate OTP verification for website signups using crypto payments.
Self Protocol Identity Verification
A privacy-preserving identity integration that uses zero-knowledge proofs to verify passports and ID cards without exposing personal data.
Etherscan EVM Data Query
An advanced interface for querying multi-chain EVM data, including balances, transactions, and smart contract details via the Etherscan API v2.
Featured*
