AirOps Integration Agent

Overview

The AirOps Integration Agent is a prototype system that enables users to configure and execute integration actions (e.g., create a Webflow item, send a Slack notification, post to WordPress) using natural language requests. It leverages workflow context and available integration actions to generate structured, executable workflows.

Features

• Natural Language Understanding: Parses user requests to extract intent, platform, entity type, and parameters.
• Action Selection: Matches parsed requests to the most relevant integration action using scoring logic.
• Schema Retrieval: Retrieves the schema for the selected action to understand required parameters.
• Parameter Generation & Validation: Extracts and validates parameters from the user request and context.
• Workflow Generation: Assembles a workflow definition, including necessary data transformations.
• Observability & Tracking: Integrates with LangSmith and Weights & Biases for tracing and performance monitoring.
• Testing: Includes scripts to test the agent with sample requests and output the generated workflows.

Project Structure

airops_integration_agent/
├── config/
│   └── observability.py
├── data/
│   ├── integration_actions.txt / .json
│   ├── workflow_context.json
│   └── sample_requests.json
├── src/
│   ├── nlp.py
│   ├── action_selector.py
│   ├── graph.py
│   ├── models/
│   │   ├── action.py
│   │   ├── request.py
│   │   └── workflow.py
│   ├── nodes/
│   │   ├── query_refiner.py
│   │   ├── planner.py
│   │   ├── schema_retriever.py
│   │   ├── generator.py
│   │   ├── validator.py
│   │   ├── repair.py
│   │   ├── workflow_generator.py
│   │   └── final_output.py
│   └── tests/
│       ├── test_integration.py
│       └── test_nlp.py
├── utils/
│   ├── helpers.py
│   └── tracking.py
├── main.py
├── requirements.txt
└── README.md

How It Works

1. User Request Parsing: The agent parses a natural language request to identify the target platform, action intent, entity type, and parameters.
2. Action Selection: It selects the most relevant integration action from available options.
3. Schema Retrieval: Retrieves the schema for the selected action to determine required parameters.
4. Parameter Generation & Validation: Extracts parameters from the request/context and validates them.
5. Workflow Generation: Generates a workflow definition, including any necessary data transformations.
6. Execution & Output: Outputs the workflow or requests clarification if needed.

Example Usage

Run the test script to see the agent in action:

python main.py

This will process sample requests and print the generated workflows and any clarification questions.

Extending the Agent

• Add new integrations: Update data/integration_actions.txt or .json with new actions.
• Improve NLP: Enhance src/nlp.py for more robust request parsing.
• Add nodes: Implement new workflow steps in src/nodes/ as needed.

Observability & Tracking

• LangSmith: Used for tracing LLM calls and agent steps.
• Weights & Biases (wandb): Used for performance tracking.

Requirements

• Python 3.10+
• See requirements.txt for dependencies.

Contributing

1. Fork the repo and create a feature branch.
2. Add or update tests in src/tests/.
3. Submit a pull request with a clear description.

License

MIT License

---

For more details, see the code and docstrings in each module.