[Enhancement] Show paste content preview in input placeholder before submission

Resolved 💬 4 comments Opened Feb 25, 2026 by mikegilchrist Closed Mar 26, 2026

Describe the feature

When multi-line content is pasted into the Claude Code input buffer, it is replaced with a placeholder such as:

[Pasted text #1 +22 lines]

This confirms that a paste occurred and how many lines were received, but provides no way to verify which content was captured. Users cannot confirm the correct text was pasted without submitting the message first.

Requested behavior: expand the placeholder to include a truncated preview of the first and last lines:

[Pasted text #1 +22 lines | "Exams/1/exam.1_2022_intro..." ... "evince Doolittle_2017...pdf &"]

Why this matters

  • Multi-line pastes are opaque: the line count alone does not confirm the right content was captured
  • Clipboard state in complex desktop environments (multiple clipboards, clipboard managers, tmux buffers) is often ambiguous -- users frequently are not certain which buffer was pasted
  • There is no way to inspect paste content before committing to submission

Current behavior

Placeholder shows only line count. User must submit to see what was received.

Expected behavior

Placeholder shows line count plus first ~60 chars of the first line and last ~60 chars of the last line, giving enough context to verify the paste without submitting.

Implementation note

Claude Code already intercepts paste events to generate the placeholder -- the content is available at that point. The proposed change is additive: extract the first and last lines from the already-captured paste buffer and include them in the placeholder string. No new mechanism is required.

Related issues

  • #28137 (paste race condition): a related paste UX gap where Enter fires before content is inserted. Both issues reflect insufficient user visibility into paste state before submission -- this request is the verification side of the same problem.
  • #27759 (pasted text not visible to subagents): pasted content visibility is already a known gap in other contexts; this request addresses the same gap at the input level.

Platform

Linux (confirmed), likely affects all platforms.

View original on GitHub ↗

This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗