# ucbepic/docetl

A system for agentic LLM-powered data processing and ETL

Repository: https://github.com/ucbepic/docetl
Canonical: https://ross.abutalabs.com/products/docetl
Homepage: https://docetl.org
Language: Python
License: MIT
License Family: permissive
Topics: data, etl, llm, python, data-pipelines, elt, workflow, agents, semantic-data, document-processing, unstructured-data, unstructured-data-analysis, document-analysis
Last push: 2026-08-10T15:00:22+00:00

## Health v2 (maintenance only)
Score: 86/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 97, release rhythm 89, longevity 56
- inputs: {"age_days": 785, "days_push": 23, "days_rel": 77, "gap_med": 22, "n_releases_24m": 14}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3995, forks 427 (observed 2026-08-28T04:08:31.870915+00:00)

## What it is
DocETL is a Python library and CLI for building LLM-powered data processing and ETL pipelines over structured and unstructured data using declarative map-reduce operations written in natural language. It automatically optimizes pipelines by swapping models, rewriting prompts, and decomposing operations to improve accuracy and reduce cost.

## Use cases
- extract structured fields from unstructured documents with llms
- classify and summarize thousands of support tickets
- build etl pipelines that use llm calls as operators
- process large collections of pdfs or text into queryable tables
- optimize llm data processing pipelines for cost and accuracy
- run map-reduce style analysis over unstructured text data

## When to choose
- you need to process large volumes of unstructured text or documents with llms
- you want declarative, optimizable llm pipelines instead of hand-wiring individual calls
- you want map-reduce style semantic operators with automatic parallelization
- you prefer python or yaml definitions for data transformation workflows

## When to avoid
- your data is already structured and needs only deterministic transformations
- you cannot send your data to external llm providers due to privacy constraints
- you need real-time low-latency processing rather than batch pipelines
- you want a full visual etl platform without any coding

## Facets
- artifact type: library
- maturity: active
- function: etl, llm-inference, agent-framework, nlp, data-science, workflow-automation, cli
- domain: large-language-models, developer-tools
- platform: python, cli, cross-platform
- tags: llm-pipelines, unstructured-data, document-processing, map-reduce, semantic-data-processing, data-pipelines, docwrangler, yaml-pipelines, data-engineering, natural-language-processing, ai-agents, automation

## Member repositories
- ucbepic/docetl (main) score 86

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:31.870915+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 2026-08-29T18:24:11.817421+00:00, confidence not recorded.
  - readme: https://github.com/ucbepic/docetl (fetched 2026-08-28T04:08:31.870915+00:00, sha 2056bba8bfa7)
  - homepage: https://docetl.org (fetched 2026-08-29T09:17:22.105050+00:00, sha e0292c304fb5)
- Data as of 2026-08-30T08:39:29.467469+00:00.