Today the ingestion API supports PDF, URL, YouTube and crawl. Plain `.txt` and `.docx` are common knowledge-base formats worth adding. - [ ] Accept `.txt` uploads (read as UTF-8) - [ ] Accept `.docx` (e.g. via the `mammoth` package) - [ ] Wire into the existing chunk → embed pipeline in `src/services/` See `src/routes/documents.js` and the chunker/embedder services for where to plug in.
Today the ingestion API supports PDF, URL, YouTube and crawl. Plain
.txtand.docxare common knowledge-base formats worth adding..txtuploads (read as UTF-8).docx(e.g. via themammothpackage)src/services/See
src/routes/documents.jsand the chunker/embedder services for where to plug in.