Devin Mind

A comprehensive workflow orchestration tool for Devin AI that enables seamless execution of multi-step workflows, playbook management, and cross-session automation.

App GUI

Why Devin Mind is Useful

🔗 Seamless Playbook Integration

Devin currently limits users to 1 playbook per session, making it cumbersome to chain multiple playbooks together. Devin Mind streamlines this by automatically connecting and executing multiple playbooks across sessions, eliminating manual handoffs.

📋 Workflow-First Approach

Instead of thinking in terms of individual sessions, Devin Mind lets you focus on the bigger picture - your complete workflow. Plan once, execute seamlessly, regardless of how many Devin sessions are required.

⚡ Automation & Efficiency

• Automatic polling and session management
• Intelligent handoff between workflow steps
• Reduced manual intervention and context switching
• Consistent execution across complex multi-step processes

🎯 Enhanced Productivity

• No more copy-pasting results between sessions
• Automatic context preservation across steps
• Parallel execution support for independent tasks
• Centralized workflow management and monitoring

🔄 Reusability & Consistency

• Define workflows once, run them multiple times
• Version control your automation processes
• Share and collaborate on workflow templates
• Standardize common development patterns

Features

• Polling Mechanism: Automatically poll sessions to detect completion
• Handoff Management: Handle step-to-step communication and result passing
• API Integration: Direct integration with Devin AI API
• Markdown Workflow Parsing: Parse and execute workflows defined in markdown files
• Cross-Session Orchestration: Seamlessly chain multiple Devin sessions
• Desktop Application: Native macOS and Windows apps for easy workflow management

Limitations

Devin Mind is designed for executing proven and repeatable workflows that have already been validated.
It is not intended for experimental or unverified workflows where human review and correction are still necessary.

Current limitations include:

• ❌ No built-in error handling for unexpected Devin responses
• ❌ No checkpoint or approval mechanism when a workflow step requires human validation
• ✅ Best suited for stable workflows that run successfully end-to-end with minimal supervision

⚠️ That said, nothing prevents you from running unverified workflows with Devin Mind.
Just keep in mind that you may need to manually step into the Devin session yourself and finish with a single message: sleep

Installation

CLI Installation

Install globally via npm:

npm install -g devin-mind

Or install locally in your project:

npm install devin-mind

Desktop Application - Build from Source

There are no pre-built downloadable binaries due to large file size.
To use the desktop app, clone this repository and build the distribution for your platform:

git clone https://github.com/your-repo/devin-mind.git
cd devin-mind
npm install

Build for All Platforms

npm run dist

Build for macOS

npm run build-mac

Build for Windows

npm run build-win

Build for Linux

npm run build-linux

The generated installers will be available in the dist/ directory after the build completes.

Setup

Environment Variables

Set your Devin API key as an environment variable on your system:

macOS/Linux:

export DEVIN_API_KEY="your-devin-api-key-here"

Windows:

set DEVIN_API_KEY=your-devin-api-key-here

Windows PowerShell:

$env:DEVIN_API_KEY="your-devin-api-key-here"

You must also set the REPO_URL environment variable to specify the repository host URL that Devin AI will use to detect and operate on repository snapshots for your sessions.
REPO_URL should be the repository host URL, not the full repository path.

Example:

• Full repository:
  dev.azure.com/{organization}/{project}/_git/my-repo-name
• REPO_URL value:
  dev.azure.com/{organization}/{project}/_git/

export REPO_URL="dev.azure.com/{organization}/{project}/_git/"

Optionally, you can set the NODE_ENV environment variable to control the runtime environment (e.g., development, production):

Example:

export NODE_ENV=development

Permanent Setup (recommended):

Add the environment variables to your shell profile:

• macOS/Linux: Add export DEVIN_API_KEY="your-key", export REPO_URL="your-repo-url", and export NODE_ENV=development to ~/.bashrc, ~/.zshrc, or ~/.profile
• Windows: Use System Properties > Environment Variables to set them permanently

Usage

Command Line Interface

# Basic usage
devin-mind --file path/to/workflow.md

# With custom options
devin-mind --file workflow.md --pollingInterval 15 --timeout 600 --verbose true

# Use mock mode for testing (no API key required)
devin-mind --file workflow.md --mock true

Programmatic Usage

import { startWorkflow } from 'devin-mind';

const workflowContent = `
# My Workflow
## Step 1
- Prompt: Create a simple function
- Handoff: Function created
`;

// Uses DEVIN_API_KEY from environment
const result = await startWorkflow(workflowContent);

// Or provide API key explicitly
const result = await startWorkflow(workflowContent, {
  apiKey: 'your-api-key',
  pollingInterval: 15,
  timeout: 300
});

Parsing a Workflow from Markdown

import { parseWorkflow } from 'devin-mind';

const fs = require('fs');
const markdown = fs.readFileSync('examples/sample-workflows.md', 'utf-8');
const steps = parseWorkflow(markdown);

Configuration Options

| Option | Default | Description | |--------|---------|-------------| | apiKey | process.env.DEVIN_API_KEY | Your Devin API key | | pollingInterval | 10 | Polling interval in seconds | | firstPollingInterval | 90 | First poll delay in seconds | | timeout | 300 | Timeout in seconds | | verbose | true | Enable detailed logging | | useMockMode | false | Use mock API for testing |

Project Structure

devin-mind/
├── package.json
├── index.js
├── src/
│   ├── index.js           # Main entry point
│   ├── server.js          # REST API server
│   ├── workflow-parser.js # Markdown workflow parser
│   ├── devin-client.js    # Devin API client
│   └── handoff-manager.js # Workflow execution orchestrator
├── examples/              # Example workflow markdown files
├── test/                  # Unit tests
└── README.MD

Examples

See the examples/ directory for sample workflow markdown files and usage scenarios.

Documentation

• Workflow Guidelines - Comprehensive guide on how to structure and format workflows, including detailed explanations of all step attributes (Playbook ID, RelyPreviousStep, Prompt, Handoff, etc.)
• App Usage Guide - How to use the Desktop app, including button functions and settings

Testing

Run all tests:

npm test

Or run a specific test file:

npm run unit-test-handoff
npm run unit-test-parser

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

License

MIT