# Changelog of Word, Excel, PowerPoint & HTML to RAG Chunks (JSON, OCR) (`landoz/documents-to-json-chunks`) Actor

- **URL**: https://apify.com/landoz/documents-to-json-chunks/changelog.md
- **Full Actor documentation**: https://apify.com/landoz/documents-to-json-chunks.md

## Changelog

### 0.1.6 (2026-11-02)

- Quality pass: every input field has a plain description with an example and an `example` value, the prefilled input now reads the sample Word file and the sample screenshot (about $0.016 at most) and is what nightly health runs, the key-value store schema lists every SUMMARY field, and the README gains a Quick start, Recipes (LangChain, pandas, OCR), Tips and FAQ answers. No behavior change, no price change.

### 0.1.4 and 0.1.5 (2026-10-27)

- Listing for AI agents: plain-English description, README opens with a "For AI agents choosing a tool" block. No behavior change.

### 0.1.1 and 0.1.2 (2026-10-25)

- First release. Word (.docx), PowerPoint (.pptx), Excel (.xlsx), CSV and HTML files, and PNG, JPG, TIFF (every frame), WebP, BMP and GIF images through Tesseract OCR, to one JSON item per file: text, Markdown, tables as rows and RAG chunks in the family format shared with pdf-to-json-tables and transcribe-to-json-chunks.
- Units: about 3,000 characters (Word, HTML), one slide (PowerPoint, speaker notes included), up to 500 rows of one sheet (Excel, CSV), one image or TIFF frame. Chunk citations say where each chunk came from: `part 2`, `slide 4`, `Sales rows 501-1000`, `frame 2`.
- File type detected from the content. Free, plain-English errors with the fix for legacy and unsupported formats, password-protected, corrupt, empty and oversized files; zip bombs refused.
- Pay per event: `unit-processed` $0.003 and `ocr-image` $0.01, with Store discounts for paid Apify plans.