heavy-file-ingestion-codex

To summarize large PDFs efficiently, convert the heavyweight files into compressed markdown or CSV artifacts before running extraction or analysis. This deterministic conversion prevents wasted model tokens by ensuring the model context only processes the generated lightweight index and extracted text.

You can extract tables from XLSX, CSV, and TSV files by converting them into compressed CSV artifacts before analysis. This deterministic extraction approach uses openpyxl to process spreadsheets, creating lightweight indexes that allow fast table extraction without slow raw file processing.

The best way to analyze DOCX and PPTX content is converting the documents to markdown artifacts first, then reading the generated index.md to guide the analysis. This index-driven reading method uses python-docx and python-pptx to avoid hitting context limits during slide structure review or text analysis.

Heavy file ingestion supports PDFs, DOCX, PPTX, XLSX, CSV, and TSV formats by converting them into compressed markdown or CSV artifacts. It applies to workflows requesting reading, summarization, extraction, or analysis, using dependencies like pdfplumber and openpyxl for deterministic format conversion.

The model fails to process large spreadsheets when deterministic conversion tooling encounters extraction errors. The system implements fallback tiers for reliability, choosing between extracted artifact analysis, improved deterministic conversion, or manual review to handle edge cases where standard spreadsheet extraction fails.

You need to read index.md first before analyzing converted document artifacts because the ingestion process is index-driven. The index directs the model to act on the extracted lightweight artifacts, ensuring efficient summarization and extraction by using model context only on the compressed output.