classifier-extractor

name	classifier-extractor
description	Extract structured entities from source code using Pydantic schemas and Instructor-powered LLM extraction. The Classify phase of the Graftline.
version	0.1.0
author	Jero (LATTICE / MARPA Design Studios)
triggers	["classify this code","extract entities","run the classifier","what's in this file","/classify"]
tools	["Read","Glob","Grep","Bash"]

Classifier Extractor

USE WHEN the user wants to extract structured metadata from source files, run entity extraction, classify code components, or understand what a codebase contains.

What It Does

Runs the Classify phase of the Graftline — extracting structured entities from source code:

1. AST Parsing — Extract functions, classes, imports, decorators via Python AST
2. Schema Extraction — Identify Pydantic models, dataclasses, TypedDicts
3. API Discovery — Find FastAPI/Flask routes, endpoint definitions
4. Dependency Mapping — Trace import graphs and external dependencies
5. LLM Extraction — Use Instructor patterns for semantic entities AST cannot capture

Usage

/classify backend/**/*.py
/classify app/api/routes.py
classify this codebase

Entity Types

Entity	Source	Fields
Function	AST	name, args, return type, docstring, decorators
Class	AST	name, bases, methods, attributes
Import	AST	module, names, alias
Route	Pattern match	method, path, handler, auth
Model	AST + LLM	name, fields, validators, relationships
Config	LLM	key, value, source, sensitivity

Output

Structured JSON with all extracted entities, suitable for documentation generation and validation.