Quick jump / scroll to specific positions in chat (e.g. jump to "my questions")

Ideas Feature Requests
1

Feature request for product/service

Cursor IDE

Describe the request

Current problem
In long agent conversations I have to scroll a lot with the mouse to find where I asked something. There’s no way to jump directly to my messages or to a given point in the thread.

What I’m asking for

  • A way to quickly jump to “my messages” / “my questions” in the current chat (e.g. a “Jump to my messages” button or a list of my messages I can click).
  • Optionally: jump to a specific message or section (e.g. by order or rough position), so I don’t have to scroll through the whole conversation.

Why it would help
I could get to my own questions and the AI’s answers much faster, especially in long threads, instead of scrolling and often missing the part I need.

Operating System (if it applies)

MacOS

3

Feature request for product/service

Chat

Describe the request

When having long conversations with the AI agent, key insights and decisions often appear somewhere in the middle of the chat (e.g., around 1000px from the top). Currently there’s no way to bookmark or anchor specific positions in a chat conversation.

Problem:

  • Long agent conversations require scrolling back and forth to find important parts
  • No way to mark “this is the key decision/insight” in a conversation
  • Cmd+F text search is the only workaround, but you need to remember exact keywords

Proposed Solution:

  • Allow users to place scroll bookmarks at specific points in a chat conversation
  • A small bookmark icon or marker in the chat gutter/margin
  • A bookmark list panel (like VS Code’s bookmark extension for code) to jump between bookmarked positions
  • Keyboard shortcut to toggle bookmark at current scroll position (e.g., Cmd+Shift+B)

Use Cases:

  1. Bookmarking the moment the agent produces a critical architectural decision
  2. Marking the start of a specific subtask in a long multi-step conversation
  3. Quickly jumping back to a reference point (e.g., error output, agreed-upon interface design)

This would significantly improve navigation in long agent sessions where context and decisions accumulate over time.