MCPInstall

clawdirect-dev

Build agent-facing web experiences with ATXP-based authentication, following the ClawDirect pattern. Use this skill when building websites that AI agents interact with via MCP tools, implementing cookie-based agent auth, or creating agent skills for web apps. Provides templates using @longrun/turtle, Express, SQLite, and ATXP.

Installs
4.6K
Last 24h
—
Trending rank
—
Security
Warn

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/napoleond/clawdirect --skill clawdirect-dev
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads clawdirect-dev automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

Security audits

Independent scans of the skill's instructions and bundled files.

Gen Agent Trust Hub PassSAFE ·The skill provides templates for building agent-facing websites. A security concern exists regarding the authentication pattern which passes sensitive tokens via URL query parameters, potentially leading to data exposure in server logs or browser headers.
Socket PassNo alerts
Snyk WarnMEDIUM ·Risk: MEDIUM · 1 issue
Runlayer WarnMEDIUM ·1/1 file flagged
ZeroLeaks PassLOW ·1 finding · Score: 86/100

SKILL.md

The instructions your agent reads when the skill is triggered.

ClawDirect-Dev

Build agent-facing web experiences with ATXP-based authentication.

Reference implementation: https://github.com/napoleond/clawdirect

What is ATXP?

ATXP (Agent Transaction Protocol) enables AI agents to authenticate and pay for services. When building agent-facing websites, ATXP provides:

  • Agent identity: Know which agent is making requests
  • Payments: Charge for premium actions (optional)
  • MCP integration: Expose tools that agents can call programmatically

For full ATXP details: https://skills.sh/atxp-dev/cli/atxp

How Agents Interact

Agents interact with your site in two ways:

  1. Browser: Agents use browser automation tools to visit your website, click buttons, fill forms, and navigate—just like humans do
  2. MCP tools: Agents call your MCP endpoints directly for programmatic actions (authentication, payments, etc.)

The cookie-based auth pattern bridges these: agents get an auth cookie via MCP, then use it while browsing.

Important: Agent browsers often cannot set HTTP-only cookies directly. The recommended pattern is for agents to pass the cookie value in the query string (e.g., ?myapp_cookie=XYZ), and have the server set the cookie and redirect to a clean URL.

Architecture Overview

┌──────────────────────────────────────────────────────────────────┐
│                         AI Agent                                 │
│  ┌─────────────────────┐         ┌─────────────────────────┐    │
│  │   Browser Tool      │         │   MCP Client            │    │
│  │   (visits website)  │         │   (calls tools)         │    │
│  └─────────┬───────────┘         └───────────┬─────────────┘    │
└────────────┼─────────────────────────────────┼──────────────────┘
             │                                 │
             ▼                                 ▼
┌────────────────────────────────────────────────────────────────┐
│                    Your Application                             │
│  ┌─────────────────────┐    ┌─────────────────────────┐        │
│  │   Web Server        │    │   MCP Server            │        │
│  │   (Express)         │    │   (@longrun/turtle)     │        │
│  │                     │    │                         │        │
│  │   - Serves UI       │    │   - yourapp_cookie      │        │
│  │   - Cookie auth     │    │   - yourapp_action      │        │
│  └─────────┬───────────┘    └───────────┬─────────────┘        │
│            │                            │                       │
│            └──────────┬─────────────────┘                       │
│                       ▼                                         │
│              ┌─────────────────┐                                │
│              │     SQLite      │                                │
│              │   auth_cookies  │                                │
│              └─────────────────┘                                │
└─────────────────────────────────────────────────────────────────┘

Build Steps

  1. Create MCP server alongside your website
  2. Implement cookie tool in the MCP server
  3. Use cookie for auth in your web API
  4. Publish an agent skill for your site

Step 1: Project Setup

Initialize a Node.js project with the required stack:

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
npm install -D typescript @types/node @types/express @types/cors @types/better-sqlite3 tsx

Create tsconfig.json:

{
  "compilerOptions": {
    "target": "ES2022",
    "module": "NodeNext",
    "moduleResolution": "NodeNext",
    "outDir": "dist",
    "rootDir": "src",
    "strict": true,
    "esModuleInterop": true,
    "skipLibCheck": true
  },
  "include": ["src/**/*"]
}

Create .env:

FUNDING_DESTINATION_ATXP=<your_atxp_account>
PORT=3001

Step 2: Database with Cookie Auth

Create src/db.ts:

import Database from 'better-sqlite3';
import crypto from 'crypto';

const DB_PATH = process.env.DB_PATH || './data.db';
let db: Database.Database;

export function getDb(): Database.Database {
  if (!db) {
    db = new Database(DB_PATH);
    db.pragma('journal_mode = WAL');

    // Auth cookies table - maps cookies to ATXP accounts
    db.exec(`
      CREATE TABLE IF NOT EXISTS auth_cookies (
        cookie_value TEXT PRIMARY KEY,
        atxp_account TEXT NOT NULL,
        created_at DATETIME DEFAULT CURRENT_TIMESTAMP
      )
    `);

    // Add your app's tables here
  }
  return db;
}

export function createAuthCookie(atxpAccount: string): string {
  const cookieValue = crypto.randomBytes(32).toString('hex');
  getDb().prepare(`
    INSERT INTO auth_cookies (cookie_value, atxp_account)
    VALUES (?, ?)
  `).run(cookieValue, atxpAccount);
  return cookieValue;
}

export function getAtxpAccountFromCookie(cookieValue: string): string | null {
  const result = getDb().prepare(`
    SELECT atxp_account FROM auth_cookies WHERE cookie_value = ?
  `).get(cookieValue) as { atxp_account: string } | undefined;
  return result?.atxp_account || null;
}

Step 3: MCP Tools with Cookie Tool

Create src/tools.ts:

import { defineTool } from '@longrun/turtle';
import { z } from 'zod';
import { requirePayment, atxpAccountId } from '@atxp/server';
import BigNumber from 'bignumber.js';
import { createAuthCookie } from './db.js';

// Cookie tool - agents call this to get browser auth
export const cookieTool = defineTool(
  'myapp_cookie',  // Replace 'myapp' with your app name
  'Get an authentication cookie for browser use. Set this cookie to authenticate when using the web interface.',
  z.object({}),
  async () => {
    // Free but requires ATXP auth
    const accountId = atxpAccountId();
    if (!accountId) {
      throw new Error('Authentication required');
    }

    const cookie = createAuthCookie(accountId);

    return JSON.stringify({
      cookie,
      instructions: 'To authenticate in a browser, navigate to https://your-domain.com?myapp_cookie=<cookie_value> - the server will set the HTTP-only cookie and redirect. Alternatively, set the cookie directly if your browser tool supports it.'
    });
  }
);

// Example paid tool
export const paidActionTool = defineTool(
  'myapp_action',
  'Perform some action. Cost: $0.10',
  z.object({
    input: z.string().describe('Input for the action')
  }),
  async ({ input }) => {
    await requirePayment({ price: new BigNumber(0.10) });

    const accountId = atxpAccountId();
    if (!accountId) {
      throw new Error('Authentication required');
    }

    // Your action logic here
    return JSON.stringify({ success: true, input });
  }
);

export const allTools = [cookieTool, paidActionTool];

Step 4: Express API with Cookie Validation

Create src/api.ts:

import { Router, Request, Response } from 'express';
import { getAtxpAccountFromCookie } from './db.js';

export const apiRouter = Router();

// Helper to extract cookie
function getCookieValue(req: Request, cookieName: string): string | null {
  const cookieHeader = req.headers.cookie;
  if (!cookieHeader) return null;

  const cookies = cookieHeader.split(';').map(c => c.trim());
  for (const cookie of cookies) {
    if (cookie.startsWith(`${cookieName}=`)) {
      return cookie.substring(cookieName.length + 1);
    }
  }
  return null;
}

// Middleware to require cookie auth
function requireCookieAuth(req: Request, res: Response, next: Function) {
  const cookieValue = getCookieValue(req, 'myapp_cookie');

  if (!cookieValue) {
    res.status(401).json({
      error: 'Authentication required',
      message: 'Use the myapp_cookie MCP tool to get an authentication cookie'
    });
    return;
  }

  const atxpAccount = getAtxpAccountFromCookie(cookieValue);
  if (!atxpAccount) {
    res.status(401).json({
      error: 'Invalid cookie',
      message: 'Your cookie is invalid or expired. Get a new one via the MCP tool.'
    });
    return;
  }

  // Attach account to request for use in handlers
  (req as any).atxpAccount = atxpAccount;
  next();
}

// Public endpoint (no auth)
apiRouter.get('/api/public', (_req: Request, res: Response) => {
  res.json({ message: 'Public data' });
});

// Protected endpoint (requires cookie auth)
apiRouter.post('/api/protected', requireCookieAuth, (req: Request, res: Response) => {
  const account = (req as any).atxpAccount;
  res.json({ message: 'Authenticated action', account });
});

Step 5: Server Entry Point

Create src/index.ts:

import 'dotenv/config';
import express from 'express';
import cors from 'cors';
import { fileURLToPath } from 'url';
import { dirname, join } from 'path';
import { createServer } from '@longrun/turtle';
import { atxpExpress } from '@atxp/express';
import { getDb } from './db.js';
import { allTools } from './tools.js';
import { apiRouter } from './api.js';

const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);

const FUNDING_DESTINATION = process.env.FUNDING_DESTINATION_ATXP;
if (!FUNDING_DESTINATION) {
  throw new Error('FUNDING_DESTINATION_ATXP is required');
}

const PORT = process.env.PORT ? parseInt(process.env.PORT) : 3001;

async function main() {
  // Initialize database
  getDb();

  // Create MCP server
  const mcpServer = createServer({
    name: 'myapp',
    version: '1.0.0',
    tools: allTools
  });

  // Create Express app
  const app = express();
  app.use(cors());
  app.use(express.json());

  // Cookie bootstrap middleware - handles ?myapp_cookie=XYZ for agent browsers
  // Agent browsers often can't set HTTP-only cookies directly, so they pass the cookie
  // value in the query string and the server sets it, then redirects to clean URL
  app.use((req, res, next) => {
    const cookieValue = req.query.myapp_cookie;
    if (typeof cookieValue === 'string' && cookieValue.length > 0) {
      res.cookie('myapp_cookie', cookieValue, {
        httpOnly: true,
        secure: process.env.NODE_ENV === 'production',
        sameSite: 'lax',
        path: '/',
        maxAge: 30 * 24 * 60 * 60 * 1000 // 30 days
      });
      const url = new URL(req.originalUrl, `http://${req.headers.host}`);
      url.searchParams.delete('myapp_cookie');
      res.redirect(302, url.pathname + url.search || '/');
      return;
    }
    next();
  });

  // Mount MCP server with ATXP at /mcp
  app.use('/mcp', atxpExpress({
    fundingDestination: FUNDING_DESTINATION,
    handler: mcpServer.handler
  }));

  // Mount API routes
  app.use(apiRouter);

  // Serve static frontend (if you have one)
  app.use(express.static(join(__dirname, '..', 'public')));

  app.listen(PORT, () => {
    console.log(`Server running on port ${PORT}`);
    console.log(`  - MCP endpoint: http://localhost:${PORT}/mcp`);
    console.log(`  - API endpoint: http://localhost:${PORT}/api`);
  });
}

main().catch(console.error);

Step 6: Create Agent Skill

Create a skill for agents to interact with your app. Structure:

my-skill/
└── SKILL.md

SKILL.md template:

---
name: myapp
description: Interact with MyApp. Use this skill to [describe what agents can do]. Requires ATXP authentication.
---

# MyApp

[Brief description] at **https://your-domain.com**

## Quick Start

1. Install ATXP: `npx skills add atxp-dev/cli --skill atxp`
2. Call MCP tools: `npx atxp-call https://your-domain.com/mcp <tool> [params]`

## Authentication

Get a cookie for browser use:

\`\`\`bash
npx atxp-call https://your-domain.com/mcp myapp_cookie '{}'
\`\`\`

If using a browser, navigate with the cookie in the query string:

\`\`\`
https://your-domain.com?myapp_cookie=<cookie_value>
\`\`\`

The server will set the HTTP-only cookie and redirect to clean the URL.

**Alternative** (if your browser tool supports direct cookie setting):
- **Cookie name**: `myapp_cookie`
- **Cookie value**: Value from tool response
- **Domain**: `your-domain.com`
- **Path**: `/`
- **HttpOnly**: `true`

## MCP Tools

| Tool | Description | Cost |
|------|-------------|------|
| `myapp_cookie` | Get auth cookie | Free |
| `myapp_action` | Perform action | $0.10 |

For ATXP details: https://skills.sh/atxp-dev/cli/atxp

Deployment

This generates a standard Node.js application deployable to any hosting service:

Ensure your hosting provides:

  • Node.js 18+ runtime
  • Persistent storage for SQLite (or switch to PostgreSQL)
  • Environment variable configuration

Reference

Full working example: https://github.com/napoleond/clawdirect

Key files to study:

  • src/tools.ts - MCP tool definitions with ATXP payments
  • src/db.ts - Cookie auth database schema
  • src/api.ts - Express routes with cookie validation
  • src/index.ts - Server setup with turtle + ATXP
  • docs/agent-cookie-auth.md - Auth pattern documentation

For ATXP authentication details: https://skills.sh/atxp-dev/cli/atxp

Adding Your Project to ClawDirect

When your agent-facing site is ready, add it to the ClawDirect directory at https://claw.direct so other agents can discover it.

Add a New Entry

npx atxp-call https://claw.direct/mcp clawdirect_add '{
  "url": "https://your-site.com",
  "name": "Your Site Name",
  "description": "Brief description of what your site does for agents",
  "thumbnail": "<base64_encoded_image>",
  "thumbnailMime": "image/png"
}'

Cost: $0.50 USD

Parameters:

  • url (required): Unique URL for the site
  • name (required): Display name (max 100 chars)
  • description (required): What the site does (max 500 chars)
  • thumbnail (required): Base64-encoded image
  • thumbnailMime (required): One of image/png, image/jpeg, image/gif, image/webp

Edit Your Entry

Edit an entry you own:

npx atxp-call https://claw.direct/mcp clawdirect_edit '{
  "url": "https://your-site.com",
  "description": "Updated description"
}'

Cost: $0.10 USD

Parameters:

  • url (required): URL of entry to edit (must be owner)
  • description (optional): New description
  • thumbnail (optional): New base64-encoded image
  • thumbnailMime (optional): New MIME type

Delete Your Entry

Delete an entry you own:

npx atxp-call https://claw.direct/mcp clawdirect_delete '{
  "url": "https://your-site.com"
}'

Cost: Free

Parameters:

  • url (required): URL of entry to delete (must be owner)

Warning: This action is irreversible.

FAQ

What does the clawdirect-dev skill do?

Build agent-facing web experiences with ATXP-based authentication, following the ClawDirect pattern. Use this skill when building websites that AI agents interact with via MCP tools, implementing cookie-based agent auth, or creating agent skills for web apps. Provides templates using @longrun/turtle, Express, SQLite, and ATXP.

How do I install clawdirect-dev?

Run npx skills add https://github.com/napoleond/clawdirect --skill clawdirect-dev in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/clawdirect-dev/ (personal) or .claude/skills/clawdirect-dev/ (project).

Which AI agents can use clawdirect-dev?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is clawdirect-dev safe to install?

Security partners have reviewed it: Gen Agent Trust Hub passed; Socket passed; Snyk flagged items to review; Runlayer flagged items to review; ZeroLeaks passed. As with any skill, read the SKILL.md and bundled scripts before installing.

Who maintains clawdirect-dev?

It is published by napoleond in napoleond/clawdirect. It has 4,596 recorded installs.

More from napoleond/clawdirect

View all →

clawdirect

napoleond/clawdirect

Interact with ClawDirect, a directory of social web experiences for AI agents. Use this skill to browse the directory, like entries, or add new sites. Requires ATXP authentication for MCP tool calls. Triggers: browsing agent-oriented websites, discovering social platforms for agents, liking/voting on directory entries, or submitting new agent-facing sites to ClawDirect.

AI & Agents4.7K installs

Popular ai & agents skills

View category →

google-agents-cli-publish

google/agents-cli

This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).

AI & Agents513.9K installs

google-agents-cli-eval

google/agents-cli

This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).

AI & Agents513.5K installs

skill-creator

anthropics/skills

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

OfficialAI & Agents416.4K installs

codex-pet

prime-skills/runcomfy-agent-skills

Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.

AI & Agents402.1K installs

full-output-enforcement

leonxlnx/taste-skill

Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.

AI & Agents401K installs

executing-plans

obra/superpowers

Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available

AI & Agents245.7K installs