> ## Documentation Index
> Fetch the complete documentation index at: https://docs.usealmanac.com/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> This handbook preview targets personal CLI 0.11.1. Check installed command help and the matching backend schema before using examples with another release. Examples with named people and placeholder IDs are fictional.

# Download original files and media

> Separate extracted content from original bytes and verify the saved file.

Reading a public PDF can extract its text, while downloading preserves the PDF bytes. A video transcript is speech text, and a watch-page URL is a page address; neither is a video file. Use the representation the task actually needs, and say which one you obtained.

This URL was used in a September 2026 read-only test. The first command returned `data.content: "Dummy PDF file\n\n"`; the second saved a one-page PDF. It illustrates two distinct operations, not a promise that every PDF extracts successfully.

```sh theme={"theme":{"light":"github-light","dark":"vesper"}}
almanac web read --source web --type pdf --engine almanac 'https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf' --output pdf-result.json
almanac web download 'https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf' --output original.pdf
```

An image or video may instead appear as an `asset` in a saved result. Inspect its `kind`, variants and source, then pass the **returned** asset ID with the result file:

```sh theme={"theme":{"light":"github-light","dark":"vesper"}}
almanac web download --from reel.json --asset instagram:0:image:0 --output reel-cover.jpg
```

That asset ID came from one September 2026 Instagram result and is not reusable for another result. The test downloaded and verified JPEG bytes; its MP4 asset was only a descriptor, so video-byte retrieval was not established. For multiple returned assets, `--all-assets --output-dir DIRECTORY` can save a batch; inspect each file and the command's receipt before saying the originals were kept. Signed media URLs can expire. The files remain on the caller's machine; a download does not register an Almanac source or attach a citation to saved knowledge. See [sources and citations](/concepts/sources-pages-and-citations) when a finding belongs in the wiki.