🛠️ `build-parser` – Interactive Pydantic Model Builder

The build-parser command launches an intuitive web-based tool to visually create Pydantic v2 models—without writing a single line of code. It's great for designing OutputParser formats, structured data schemas, or simply organizing typed data models.

🚀 Getting Started

Make sure your Python package is installed in an environment where the CLI is available. Then, simply run:

build-parser

This command launches a local Streamlit app in your browser.

📦 OutputParser Root Model

At the core of this tool is the OutputParser class. This is the primary model used by the LLM in your framework to parse outputs into structured JSON.

You always start with the OutputParser model, and build its fields just like any Pydantic model. You can then create additional models to use as nested types or components within OutputParser.

This approach supports building complex, hierarchical schemas using composition and reuse.

🧱 Interface Overview

The Model Manager allows you to create additional models beyond OutputParser.

Why is this useful?

Often, structured outputs require nested objects or repeated elements. Instead of defining everything flatly in OutputParser, you can:

Break down your schema into logical subcomponents
Reuse those components across multiple fields
Keep your data models clean, readable, and maintainable

Example:

You could define a Product model and use it as a field in OutputParser like:

class OutputParser(BaseModel):
    products: list[Product]

How to use it:

Enter a model name (e.g., Product, Metadata)
Click ➕ Add model
Names must begin with a capital letter and be valid Python identifiers

🏗️ Design Tab

Use this section to add fields to OutputParser and any other models you've created.

Field name
Field type (primitive, special type, or reference to another model)
Mark as Optional if needed
Provide subtypes for list/dict, or values for Literal

📝 Code Tab

This shows the full generated code using Pydantic v2:

Starts with the OutputParser model
Includes any additional models you’ve defined

💾 Export Tab

Choose to either:

Download the generated code as a .py file
Or save directly to tasks/parsers/ inside your project

🧪 Example

Let’s say you want to extract a list of products from LLM output:

Define a model called Product with fields like:
name: str
price: float
In OutputParser, add:
products: list[Product]

Resulting code:

from pydantic import BaseModel

class Product(BaseModel):
    name: str
    price: float

class OutputParser(BaseModel):
    products: list[Product]

📁 Output Location

If you choose "Save to tasks/parsers/", your file will be saved at:

<project-root>/tasks/parsers/<filename>.py

🛠️ build-parser – Interactive Pydantic Model Builder