You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[FEATURE]: Make paste summary thresholds configurable #15767
I have verified this feature I'm about to request hasn't been suggested before.
(I did submit it in #14712 but that got auto-closed before I noticed for not having [FEATURE]: in the title.)
Describe the enhancement you want to request
The paste summary feature is genuinely useful when pasting large chunks of content — it keeps the prompt input clean and manageable. However, the current thresholds (3 lines / 150 characters) are too aggressive for those of us who use voice dictation software (e.g. Whisper Flow, Whispering, others) on a regular basis.
Dictated text frequently exceeds 150 characters in a single utterance, causing it to be immediately collapsed into [Pasted ~1 lines]. This makes it impossible to spot and correct the small transcription errors that dictation software occasionally makes — which is exactly the moment you most want to see what was pasted. This is the scenario described in #3949, where the only workaround offered was experimental.disable_paste_summary: true — a blunt hammer that trades one problem for another: you lose paste summaries entirely, which is painful when you then paste a 200-line config file or a stack trace.
What is needed is the ability to tune the thresholds to match your typical dictation length, so paste summaries kick in only when the content is genuinely large.
Both would be optional, defaulting to the current values (3 and 150 respectively), so existing behaviour is preserved for anyone who doesn't set them.
Prior art
PR #1960 attempted this with a nested config structure (experimental.editor.summary_thresholds); this proposal uses flat keys consistent with the existing experimental config style.
Reacted by Frederick Fogerty, Chaskin Saroff, Roy Ivy III, Philip Mateescu, Matthieu Delarue and Chedi
My workaround for this right now is when I'm going to input a large prompt via voice in one go, I'll first open an editor (using /editor) and I will compose my prompt there. It also honestly gives me more control about the structure of the prompt and what I'm also pasting in, etc. But of course, doing this in open code would be a massive improvement.
For anyone still looking for this on v1: #54458 adds prompt.paste_summary to tui.json (line and character thresholds, and a placeholder format with {lines}, {characters}, {bytes} and an {excerpt} of the paste). #54459 adds a separate key, <leader>v by default, that pastes as typed without a placeholder.
Activity
github-actions commentedon Mar 2, 2026
This issue might be a duplicate of existing issues. Please check:
aspiers commentedon Mar 2, 2026
Solved by #15771.
frederickfogerty commentedon Mar 19, 2026
My workaround for this right now is when I'm going to input a large prompt via voice in one go, I'll first open an editor (using
/editor) and I will compose my prompt there. It also honestly gives me more control about the structure of the prompt and what I'm also pasting in, etc. But of course, doing this in open code would be a massive improvement.github-actions commentedon Jul 3, 2026
To stay organized issues are automatically closed after 60 days of no activity. If the issue is still relevant please open a new one.
Nowaker commentedon Oct 11, 2026
For anyone still looking for this on v1: #54458 adds
prompt.paste_summarytotui.json(line and character thresholds, and a placeholder format with{lines},{characters},{bytes}and an{excerpt}of the paste). #54459 adds a separate key,<leader>vby default, that pastes as typed without a placeholder.