A development framework for building AI agent-friendly web applications featuring secure ATXP authentication and MCP tool integration.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install clawdirect-dev
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 clawdirect-dev using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
ClawDirect-Dev is a comprehensive development pattern and toolkit designed for engineers building web applications that AI agents interact with. It leverages the Agent Transaction Protocol (ATXP) and Model Context Protocol (MCP) to bridge the gap between human-centric web interfaces and programmatic agent access. By utilizing Openclaw Skills, developers can implement secure, cookie-based authentication that allows agents to move seamlessly between browser-based automation and direct API tool calls.
The framework addresses a critical technical hurdle: the inability of many agent browsers to set HTTP-only cookies directly. ClawDirect-Dev solves this by providing templates for a query-string-to-cookie bootstrap pattern, ensuring agents can maintain stateful sessions. Whether you are building a niche tool or a complex platform, this skill provides the architecture needed to make your site discoverable and usable by the next generation of autonomous AI agents.
To begin building with this pattern in the context of Openclaw Skills, initialize your Node.js environment and install the core dependencies:
mkdir my-agent-app && cd my-agent-app
npm init -y
npm install @longrun/turtle @atxp/server @atxp/express better-sqlite3 express cors dotenv zod
Ensure you have a .env file configured with your ATXP destination to handle identity and payments:
FUNDING_DESTINATION_ATXP=<your_atxp_account>
PORT=3001
The architecture relies on a persistent SQLite backend to manage the relationship between agents and active sessions. The following table structure is standard:
| Table Name | Column | Type | Description |
|---|---|---|---|
auth_cookies |
cookie_value |
TEXT (PK) | The unique hex-encoded session identifier. |
auth_cookies |
atxp_account |
TEXT | The verified ATXP account ID of the agent. |
auth_cookies |
created_at |
DATETIME | Automatic timestamp for session tracking and expiry. |
All tools are defined using Zod schemas to ensure strict data validation between the MCP client and the Express backend.
Loading
A secure CLI tool designed to register AI agents and publish Markdown-formatted progress logs to the moltlog.ai platform.

A decentralized coordination protocol for AI agents using cryptographic identities, trust graphs, and consensus-driven code evolution.

SimpleMem provides efficient lifelong memory for LLM agents through semantic compression and intent-aware retrieval.

A high-performance text extraction tool designed to bridge the gap between static PDF documents and AI-ready data structures.

A specialized directory and protocol that allows AI agents to discover, interact with, and submit agent-oriented social web experiences.

A dedicated social media platform where AI agents generate, share, and interact with image-based content using the Model Context Protocol.








































