Chunkr
An open-source, self-hostable document extraction API that turns PDFs into semantically structured Markdown blocks.
Open-source document ingestion API that segments PDFs and scanned files into semantically typed chunks with bounding boxes, OCR, and Markdown output, designed for developers building retrieval pipelines who need self-hostable infrastructure.
Key Features
- Document segmentation
- Semantic type recognition
- Bounding-box localization
- Optical character recognition
- Markdown output
Pros
- Supports self-hosting to ensure privacy
- Designed for retrieval pipelines
- Structurally complete output format
Cons
- Requires development ability to integrate
- Depends on infrastructure compute resources
Use Cases
- Building enterprise knowledge-base retrieval
- Digitizing scanned files
- Preprocessing complex PDF documents
Editor's Note
An open-source parsing tool designed for RAG pipelines, with high self-hosting flexibility and precise semantic chunking.
FAQ
What kind of tool is Chunkr?
It is an open-source document extraction API designed for developers that turns PDFs and scanned files into structured Markdown chunks.
Does this tool support self-hosting?
Yes. Chunkr offers self-hostable infrastructure, suited to development teams that value data privacy and autonomy.
What traditional PDF-parsing pain points does it solve?
It prevents text fragments from losing context and accurately recognizes tables and multi-column layouts, providing high-quality output with bounding boxes and semantic types.