open-chatgpt-atlas

🥈Silver

Open-source alternative to ChatGPT Atlas. Enables operations teams to automate browser-based tasks using AI agents. Integrates with Chromium and Electron.js for desktop automation. Supports Claude and other AI models.

407660Updated 1w ago

Intermediate30min to implementautomation

Saves ~120 min per use

Quick InstallView Source

git clone https://github.com/ComposioHQ/open-chatgpt-atlas.git

Works with:

Claude

Overview

About This Skill

https://github.com/composiohq/open-chatgpt-atlas

How to Use

[{"step":1,"action":"Install open-chatgpt-atlas via npm: `npm install -g open-chatgpt-atlas`. Ensure you have Node.js (v18+) and Chromium/Electron.js dependencies installed.","tip":"Use `npx open-chatgpt-atlas --version` to verify installation. For Electron.js, add `--electron` flag."},{"step":2,"action":"Write a task configuration file (e.g., `task.json`) with the steps you want the AI agent to perform. Include placeholders like [WEBSITE_URL] and [TASK_DESCRIPTION].","tip":"Use the example output as a template. Define clear error handling in the config, such as timeouts or retry policies."},{"step":3,"action":"Run the task using: `open-chatgpt-atlas --task task.json --model claude-3-opus`. Replace `--model` with your preferred AI model (e.g., gpt-4, mistral).","tip":"For debugging, use `--verbose` to log real-time browser interactions. Monitor memory usage with `--max-memory 512`."},{"step":4,"action":"Review the generated JSON report. If errors occur, adjust the task config or AI model instructions and rerun.","tip":"For persistent issues, enable headless mode (`--headless false`) to visually inspect the browser actions."},{"step":5,"action":"Integrate the output into your workflow. For example, pipe the CSV file to a data pipeline or trigger follow-up actions in your CRM.","tip":"Use `jq` to parse the JSON output: `open-chatgpt-atlas --task task.json | jq '.steps_completed[].output'`."}]

Use Cases

Automate checking unread messages in Gmail with a simple command.

Create and manage GitHub issues directly through chat prompts.

Send messages to Slack channels without manual input.

Navigate to websites and perform actions like scrolling or clicking with natural language commands.

Setup & Installation

Quick Install

No install command available. Check the GitHub repository for manual installation instructions.

Alternative Install (Git Clone)

git clone https://github.com/ComposioHQ/open-chatgpt-atlas

Requirements

Claude Code or compatible AI agent
Works with: Claude

Quick Start Guide

Install the Skill

Copy the install command above and run it in your terminal.

Open Your AI Agent

Launch Claude Code, Cursor, or your preferred AI coding agent.

Try It Out

Use the prompt template or examples below to test the skill.

Customize

Adapt the skill to your specific use case and workflow.

Usage Examples

Prompt Template

Automate the following browser-based task using open-chatgpt-atlas: [TASK_DESCRIPTION]. Use Chromium/Electron.js to interact with [WEBSITE_URL]. Ensure the AI agent follows these steps: [STEP_1], [STEP_2], and [STEP_3]. Handle errors like [COMMON_ERRORS] gracefully. Return a structured JSON report with [REQUIRED_FIELDS].

Example Output

```json
{
  "task_summary": "Automated login and data extraction from the client portal of Acme Corp.",
  "steps_completed": [
    {
      "step": 1,
      "action": "Navigated to https://portal.acmecorp.com",
      "status": "success",
      "timestamp": "2023-11-15T14:30:22Z",
      "output": "Login page loaded successfully"
    },
    {
      "step": 2,
      "action": "Entered username 'ops_team@acmecorp.com' and password 'SecurePass123!'",
      "status": "success",
      "timestamp": "2023-11-15T14:30:45Z",
      "output": "Authentication successful, redirected to dashboard"
    },
    {
      "step": 3,
      "action": "Clicked on 'Reports' tab and exported CSV for Q3 2023",
      "status": "success",
      "timestamp": "2023-11-15T14:31:12Z",
      "output": {
        "file_path": "/downloads/acme_q3_2023.csv",
        "rows_extracted": 1247,
        "columns": ["ClientID", "Revenue", "Status", "LastUpdated"]
      }
    }
  ],
  "errors_encountered": [],
  "performance_metrics": {
    "total_time_seconds": 50,
    "memory_usage_mb": 45.2,
    "cpu_usage_percent": 12.5
  },
  "recommendations": [
    "Optimize the CSV export process to reduce file size by 15%",
    "Add a retry mechanism for failed authentication attempts"
  ]
}
```