AIDAG - AI-Driven Directed Acyclic Graph Executor

AIDAG is a TypeScript library for executing AI-driven workflows defined as Directed Acyclic Graphs (DAGs). It supports various node types, including AI text generation and HTTP requests, allowing for complex, multi-step AI workflows.

Features

• Define workflows as Directed Acyclic Graphs (DAGs)
• Support for AI nodes (using OpenAI's GPT models)
• Support for HTTP request nodes
• Extensible architecture for adding new node types
• Asynchronous execution of workflows

Installation

Clone the repository and install the dependencies:

git clone https://github.com/govindpuff/aidag.git
cd aidag
npm install

Usage

Define a Workflow

Create a JSON file that defines your workflow. Here's an example of a simple workflow that fetches a dad joke and translates it to French:

{
  "nodes": {
    "fetchJoke": {
      "id": "fetchJoke",
      "type": "HTTP",
      "params": {
        "url": "https://icanhazdadjoke.com",
        "method": "GET",
        "outputPath": "joke",
        "headers": {
          "Accept": "application/json"
        }
      },
      "children": ["translateJoke"]
    },
    "translateJoke": {
      "id": "translateJoke",
      "type": "AI",
      "params": {
        "prompt": "Translate the following joke to French:\n\n{input}\n\nFrench translation:",
        "model": "gpt-3.5-turbo",
        "temperature": 0.7,
        "max_tokens": 200
      },
      "children": []
    }
  },
  "rootNodes": ["fetchJoke"]
}

Run the Workflow

To run a workflow, use the executeWorkflow function. You can create a script like examples/runner.ts to execute your workflows:

import { executeWorkflow } from "../src/index"
import fs from "fs"

const runWorkflow = async (filePath: string) => {
  const workflowJson = fs.readFileSync(filePath, "utf-8")
  const workflow = JSON.parse(workflowJson)

  const result = await executeWorkflow(workflow, "", { debug: true })
  console.log(JSON.stringify(result, null, 2))
}

const filePath = process.argv[2]
if (!filePath) {
  console.error("Please provide a workflow JSON file path")
  process.exit(1)
}

runWorkflow(filePath).catch(console.error)

Then, you can run your workflow using ts-node:

npx ts-node examples/runner.ts examples/dad-joke-translation.json

Make sure to set the OPENAI_API_KEY environment variable with your OpenAI API key before running the script.

Extending AIDAG

Adding New Node Types

To add a new node type:

1. Create a new class that implements the DAGNode interface in src/nodes/.
2. Implement the execute method for the new node type.
3. Add a new case in the DAGNodeFactory to handle the new node type.
4. Update the types.ts file to include any new interfaces or types needed for the new node.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License.