# deepauto-ai/automl-agent architecture

> Architecture diagram of the GitHub repository deepauto-ai/automl-agent, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2025-07-30). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/deepauto-ai/automl-agent
- Repository: https://github.com/deepauto-ai/automl-agent

## Overview

##### Project type and purpose
- AutoML-Agent is a Python-based multi-agent framework that automates the full AutoML pipeline via LLMs (ICML 2025).
- It orchestrates several specialized agents (Prompt, Data, Model, Operation) under a central AgentManager and interacts with external LLM services (vLLM or OpenAI).

##### File structure
- Top-level modules:
- agent_manager – core orchestrator (AgentManager, chat loop)
- prompt_agent – handles prompt generation & parsing (WizardLAMP templates, schemas)
- data_agent – inspects data, infers schema, creates preprocessing plans
- model_agent – selects/trains ML models, tunes hyperparameters
- operation_agent – executes code (e.g., training, evaluation)
- prompt_pool – curated prompts per task modality (image, text, tabular, graph, time series)
- experiments – scripts for evaluation, constraint-based or free-form prompting
- Static assets (index.html + static/) suggest a lightweight web interface; example_plans/ provides plan templates.
- configs.py holds LLM backends and API endpoints. requirements.txt pins dependencies.

##### README for insights
- Describes multi‐agent design, external LLM servers (vLLM or OpenAI), data flow (upload CSV → AgentManager → agents → generated code in agent_workspace).

## Diagram (Mermaid)

```mermaid
flowchart TB
    %% UI
    UI["User Interface<br/>(Jupyter/CLI & Web UI)"]:::ui
    click UI "https://github.com/deepauto-ai/automl-agent/blob/main/index.html"

    %% Orchestration
    AgentManager["AgentManager"]:::orchestration
    click AgentManager "https://github.com/deepauto-ai/automl-agent/tree/main/agent_manager/"

    UI -->|"submit data & prompt"| AgentManager

    %% Agents
    subgraph "Agents" 
        direction TB
        PromptAgent["Prompt Agent"]:::agent
        DataAgent["Data Agent"]:::agent
        ModelAgent["Model Agent"]:::agent
        OperationAgent["Operation Agent"]:::agent
    end

    click PromptAgent "https://github.com/deepauto-ai/automl-agent/tree/main/prompt_agent/"
    click DataAgent "https://github.com/deepauto-ai/automl-agent/tree/main/data_agent/"
    click ModelAgent "https://github.com/deepauto-ai/automl-agent/tree/main/model_agent/"
    click OperationAgent "https://github.com/deepauto-ai/automl-agent/tree/main/operation_agent/"

    AgentManager -->|"plan decomposition"| PromptAgent
    AgentManager -->|"analyze dataset"| DataAgent
    AgentManager -->|"select & configure model"| ModelAgent
    AgentManager -->|"execute code"| OperationAgent

    %% LLM Backends
    subgraph "LLM Backends"
        direction TB
        vLLM["vLLM Server"]:::external
        OpenAI["OpenAI API"]:::external
    end

    PromptAgent -->|"prompt →"| vLLM
    PromptAgent -->|"prompt →"| OpenAI
    vLLM -->|"response ←"| PromptAgent
    OpenAI -->|"response ←"| PromptAgent

    %% Shared Resources
    subgraph "Shared Libraries & Resources"
        direction TB
        PromptPool["Prompt Pool"]:::resource
        ExamplePlans["Example Plans"]:::resource
        Utils["Utils"]:::resource
        Configs["configs.py"]:::resource
    end

    click PromptPool "https://github.com/deepauto-ai/automl-agent/tree/main/prompt_pool/"
    click ExamplePlans "https://github.com/deepauto-ai/automl-agent/tree/main/example_plans/"
    click Utils "https://github.com/deepauto-ai/automl-agent/tree/main/utils/"
    click Configs "https://github.com/deepauto-ai/automl-agent/blob/main/configs.py"

    PromptAgent --> PromptPool
    DataAgent --> PromptPool
    ModelAgent --> PromptPool
    OperationAgent --> PromptPool

    PromptAgent --> ExamplePlans

    AgentManager --> Utils
    PromptAgent --> Utils
    DataAgent --> Utils
    ModelAgent --> Utils
    OperationAgent --> Utils

    AgentManager --> Configs
    PromptAgent --> Configs

    %% Data Storage
    AgentWorkspace["agent_workspace<br/>(datasets & scripts)"]:::storage
    OperationAgent -->|"writes code & results"| AgentWorkspace

    %% Experiments & Evaluation
    Experiments["Experiments & Evaluation"]:::resource
    click Experiments "https://github.com/deepauto-ai/automl-agent/tree/main/experiments/"

    Experiments --> AgentManager
    Experiments --> PromptAgent
    Experiments --> DataAgent
    Experiments --> ModelAgent
    Experiments --> OperationAgent

    %% Styles
    classDef orchestration fill:#AED6F1,stroke:#3498DB,color:#1B4F72
    classDef agent fill:#ABEBC6,stroke:#27AE60,color:#145A32
    classDef external fill:#F5CBA7,stroke:#E67E22,color:#6E2C00
    classDef resource fill:#E5E7E9,stroke:#A6ACAF,color:#424949
    classDef ui fill:#D6EAF8,stroke:#5DADE2,color:#154360
    classDef storage fill:#D5F5E3,stroke:#52BE80,color:#145A32
```
