# katanaml/sparrow

Structured data extraction, instruction calling and agentic workflows with ML, LLM and Vision LLM

Repository: https://github.com/katanaml/sparrow
Canonical: https://ross.abutalabs.com/products/sparrow
Homepage: https://sparrow.katanaml.io
Language: Python
License: GPL-3.0
License Family: copyleft
Topics: machinelearning, huggingface-transformers, computer-vision, llm, vllm, agentic-ai, documentai
Last push: 2026-08-26T12:04:45+00:00

## Health v2 (maintenance only)
Score: 95/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 87, longevity 100
- inputs: {"age_days": 1698, "days_push": 7, "days_rel": 89, "gap_med": 24.5, "n_releases_24m": 13}
- flags: no_readme
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5202, forks 519 (observed 2026-08-28T04:09:12.124326+00:00)

## What it is
Sparrow is an open-source framework for structured data extraction from documents (PDFs, images) using ML, LLMs, and Vision LLMs, with schema-driven extraction and agentic workflows. It combines Vision LLM extraction with Text LLM reasoning for classification, validation, and orchestration without cloud dependencies.

## Use cases
- extract structured data from invoices and receipts
- parse tables from financial reports and bank statements
- convert PDF documents to JSON with a defined schema
- classify and extract fields from multi-page documents
- build agentic document processing workflows with LLMs
- validate extracted document values against typed schemas

## When to choose
- you need schema-driven structured extraction from PDFs or document images
- you want self-hosted document AI without cloud dependencies
- you need to combine vision LLM extraction with LLM reasoning and validation

## When to avoid
- you need a simple turnkey OCR API with no ML setup
- your documents are plain text with no layout or tables
- you require a permissive license for proprietary redistribution (GPL-3.0)

## Facets
- artifact type: framework
- maturity: active
- function: machine-learning, ocr, nlp, llm-inference, rag, agent-framework, computer-vision, pdf
- domain: artificial-intelligence, machine-learning, computer-vision, pdf
- platform: python, self-hosted
- tags: document-extraction, vision-llm, structured-data, agentic-workflows, schema-validation, document-ai, natural-language-processing, documentai, docker

## Member repositories
- katanaml/sparrow (main) score 95

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:12.124326+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:00:14.834355+00:00, confidence not recorded.
  - homepage: https://sparrow.katanaml.io (fetched 2026-08-29T08:55:25.639157+00:00, sha e6bb03754cf2)
- Data as of 2026-08-30T08:39:29.467469+00:00.