> For the complete documentation index, see [llms.txt](https://docs.katara.ai/about/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.katara.ai/about/brains/sources.md).

# Sources and uploads

Sources are accepted originals uploaded directly to a Brain. Their ingestion origin determines membership: existing corpus documents cannot be attached, and changing collection filters or tags does not move a document into a Brain.

## Before uploading

Select the correct organization and Brain. Uploading requires a ready, enabled Brain and current contribution permission. The uploader must also be able to show the Brain's contribution disclosure, including audience and export exposure. If that information cannot be verified, do not assume the destination is private.

Supported extensions are `.pdf`, `.gif`, `.tif`, `.tiff`, `.jpg`, `.jpeg`, `.png`, `.bmp`, `.webp`, `.html`, `.docx`, `.pptx`, `.xlsx`, `.md`, `.markdown` and `.txt`. Each file must be no larger than **20 MiB**.

## Upload files

1. Open the Brain's **Sources** tab and select **Upload sources**.
2. Confirm the upload destination is the intended Brain, not a Content Uploader or another Brain.
3. Review the audience and export disclosure before contributing sensitive material.
4. Select your files. Review the list and remove anything you do not intend to submit.
5. Start the upload and inspect each file's outcome. Use the offered per-file retry for a failed upload rather than starting a duplicate batch.
6. Return to Sources and check which originals were accepted.

The uploader processes multiple files with up to three concurrent workers. An **accepted for processing** result is not confirmation that extraction, search indexing or Wiki generation has finished. Pending or failed first uploads are not accepted sources.

## Inspect accepted sources

Search operates on loaded sources. Load more entries if the document is not present in the current list. File and uploader information are shown when reported; missing attribution is not evidence that the current user uploaded the document.

Open a source's information sheet to inspect its accepted original content. This is a text-content view, not a promise of original binary download or a PDF preview. Per-file processing stages, keyword indexing, semantic indexing and Wiki inclusion are not reported as a complete processing pipeline.

Use **Open Content Management** for document management. The link carries this Brain's ingestion-origin filter, which remains across tabs and additional pages. **Clear origin filter** broadens that view; it does not grant access to otherwise restricted documents. See [Document management](https://docs.katara.ai/about/documents-and-collections/documents.md).

## Replace, delete or share an original

Use the document actions available under your current permissions. Original text replacement accepts plain text or Markdown. A successful replacement makes derived Wiki knowledge stale; a failed replacement does not replace the accepted original. Check [Activity](https://docs.katara.ai/about/brains/activity.md) for subsequent refreshes.

Deleting an original is not historical erasure. Derived knowledge, earlier Wiki edits and external exports may retain material from it. Review [GitHub export](https://docs.katara.ai/about/brains/github-export.md) before relying on deletion to remove disclosed information.

Brain sources inherit Brain access and may also have **Additional document viewers**. Only the Brain owner manages this source-sharing audience; the displayed uploader is not necessarily the owner. Adding a document viewer does not grant Brain, Wiki or fact access. Removing a direct viewer grant does not remove any access inherited from the Brain.

## Troubleshooting

* **Upload blocked:** check readiness, enabled state, contribution permission and disclosure loading. Organization administration alone is not sufficient.
* **Accepted file not immediately visible:** distinguish an upload receipt from an accepted source, then reload the list or load more entries.
* **Wiki still shows old content:** inspect refresh status and whether the page is authored. [Authored narrative](https://docs.katara.ai/about/brains/knowledge.md) is manually maintained.
* **Sharing appears ineffective:** check inherited access separately from additional viewers. Manage the Brain audience in [Settings](https://docs.katara.ai/about/brains/settings.md).
