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

# Fragments

> The excerpts you mark as significant, and the annotations you make.

Fragments are the excerpts you mark as significant from your Artefacts — passages of text, images (a figure, diagram, or page), or clips from audio and video — along with the annotations you make (margin-notes and other commentary) on them.

<Frame>
  <img src="https://mintcdn.com/ladderresearch/IEhuYvcofe_2f4fk/images/artefact%20reader%20view.png?fit=max&auto=format&n=IEhuYvcofe_2f4fk&q=85&s=e4c8ec3b8f0c6483a92ca3825dc01d98" alt="Artefact Reader View" width="2668" height="1876" data-path="images/artefact reader view.png" />
</Frame>

## Creating Fragments

### Ask Pond to suggest passages

Open an Artefact and choose **Suggest passages**. Pond reads the whole document and proposes passages worth marking, in two groups: those that **connect to your threads** and those **worth marking beyond them**. Accept the ones that resonate — accepting marks them to your library — and dismiss the rest. This works on any Artefact with body text: articles, notes, and web pages as well as PDFs, annotated or not, so you don't have to highlight by hand to get started.

Letting Pond suggest passages is a [membership](/pricing) feature — it's one of the agentic runs Pond performs on your behalf. Marking is still yours to do: Pond proposes, you decide what to keep.

### From PDF highlights and notes

When you upload a PDF you've already highlighted or annotated, Pond detects those marks and turns them into Fragments automatically.

### By hand, in the Artefact detail view

You can also mark Fragments yourself by selecting text in the expanded Artefact detail view:

<img src="https://mintcdn.com/ladderresearch/7KJ1UEKJ03Qtrhwx/images/create-fragment.jpg?fit=max&auto=format&n=7KJ1UEKJ03Qtrhwx&q=85&s=84312dc9553767483ddc16bc203b5798" alt="Creating a Fragment by highlighting text" width="855" height="683" data-path="images/create-fragment.jpg" />

### Image Fragments

When an Artefact is an image — or contains images, like a PDF page or figure — you can mark the image itself (or a region of it) as a Fragment. Image Fragments behave like text Fragments: you can connect them, add them to Threads, and write annotations on them.

<Note>
  Marking is yours to do. It matters that we identify for ourselves what catches our interest in a text, so Pond proposes candidates rather than marking on your behalf — nothing enters your library until you accept it. [**Frames**](/frames) work the same way, interpreting what you read through the lens of the Threads you are developing and offering ways into the work, without deciding for you.
</Note>

## Working with Fragments

Fragments are not just storage — they're meant to be moved, related, and written through. Once you have one, you can:

* **Connect** it to other Fragments via typed [Connections](/connections).
* **Add** it to a [Thread](/threads).
* **Reference** it inside a [Note](/notes) via `@`-mention (scoped to the Thread's contents). When two or more Fragments are referenced in the same Note, an inline chip suggests you connect them.