### Preflight Checklist - [x] I have searched [existing requests](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20label%3Aenhancement) and this feature hasn't been requested yet - [x] This is a single feature request (not multiple features) ### Problem Statement ## Problem When executing Bash commands that generate long outputs (>200-300 lines), Claude Code automatically collapses the output, requiring users to press Control+O to expand and view the full results. ## Cu

[FEATURE] Feature Request: Allow configuration to prevent auto-collapsing of long Bash outputs

Status Fixed / completed

Maintainer reply None cached

Workaround ✓ Mentioned in thread ↓

Activity 15 comments · opened Nov 10, 2025 · closed Aug 17, 2026

Preflight Checklist

[x] I have searched existing requests and this feature hasn't been requested yet
[x] This is a single feature request (not multiple features)

Problem Statement

Problem

When executing Bash commands that generate long outputs (>200-300 lines), Claude Code automatically collapses the output,
requiring users to press Control+O to expand and view the full results.

## Current Behavior

Long Bash outputs are automatically collapsed
User must manually expand with Control+O to see full content
No configuration option to disable this behavior

## Expected/Desired Behavior
Users should have the option to:

Configure the line threshold before auto-collapsing (or disable it entirely)
See full outputs directly without manual expansion for specific commands
Perhaps a setting like "bash.autoCollapseThreshold": 1000 in settings

## Use Case
I have a Python script that analyzes stock options data and outputs ~350 lines of formatted analysis (tables, charts,
metrics). The workflow involves:

Running analysis multiple times per day
Reviewing full output immediately each time
Currently requiring Control+O every single execution

Having to expand outputs manually on every run adds friction to an otherwise smooth workflow.

## Suggested Solution
Add a configuration option in Claude Code settings:
```json
{
"bash.autoCollapseThreshold": 1000, // or -1 to disable
"bash.alwaysExpand": false
}

This would give users control over when outputs collapse, improving workflows for data analysis, logging, testing, and
other use cases with long outputs.

Thanks for considering this enhancement!

Proposed Solution

Add a configuration option in Claude Code settings to allow users to control when Bash outputs auto-collapse:

```json
{
"bash.autoCollapseThreshold": 1000, // Number of lines before collapsing, or -1 to disable
"bash.alwaysExpand": false // Always show full output without collapsing
}

This would give users control over the auto-collapse behavior, improving workflows for data analysis, logging, testing,
and other use cases with long outputs.

Alternative: Add a per-command flag or comment syntax like # @no-collapse to prevent specific commands from being
collapsed.

Alternative Solutions

_No response_

Priority

Critical - Blocking my work

Feature Category

CLI commands and flags

Use Case Example

_No response_

Additional Context

_No response_

View original on GitHub ↗

14 Comments

databrecht · 9 months ago

+1 on this. Same friction here.

Collapsing breaks the flow, especially in interactive workflows where you want to glance at output and continue. Expanding jumps you out of the conversation.

One workaround is convincing Claude to reprint command output in its own response, but it takes some effort to get it to do this consistently, and then you lose things markup can't express anyway (colors, alignment, terminal formatting). My tool is was an idea to do things that markup can't express but I quickly ran into the same UX problems as the reporter.

github-actions[bot] · 8 months ago

This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.

cowwoc · 7 months ago

Still relevant

quidmonkey · 7 months ago

I closed my issue in favor of this one. I built a skill to show terminal dataviz when analyzing data with Claude. Unfortunately, any chart will be auto-collapsed, which is not an ideal UX for this use case.

pablomoli · 6 months ago

I'd like a settings.json option like "verboseOutput": true that persists across sessions so I don't have to press Ctrl+O every time. My use case: monitoring agent tool calls in real time during development.

olejorgenb · 6 months ago

I started using Claude Code after collapsed output was released. While I haven't experienced the full-output version, I'm quite certain I would prefer the current version.

BUT: It is super useful - often in surprising ways - to get make mini tools that Claude can call. In those cases it's better to just get the output directly than to figure out some hacky way for the model to write the output as a normal message.

Example: I instructed Claude to trace a call path from A to B and use a small script to render it nicely AND hyperlink the the output so I can jump directly to the code in my editor. This worked great, but (1) I'm unable to get it to output the links unless it's rendered as the actual tool output, and (2) it feels like waste of tokens to make it repeat the output as a message, and (3) even if (2) was not a problem it's still extra clutter showing both the toolcall and the message.

So I think having the toolcall take an argument collapsed would open up for a ton of cool ideas. Then I could just tell Claude to use collapsed=false. This is similar to what #27484 proposed (buried as a duplicate of a duplicate of this issue)

TLDR: please make this an tool argument, instead/in-addition-to a global setting

adam-stone-zocdoc · 6 months ago

+1 — Running dbt build commands that produce 25+ model/test results throughout the day. Having to click into every command to open the scrollable Shell details view adds friction to the workflow.

Would love either:

A configurable line threshold before collapsing (e.g., bash.autoCollapseThreshold)
Auto-expand into the Shell details view for long-running commands
Or even a per-command pattern match (e.g., always expand dbt * commands)

The Shell details view (scrollable, shows full output, status, runtime) is great — it just needs a way to open automatically.

sindrijo · 5 months ago

I made a mini skill to convert urls to qrcodes printed in terminal chars, it would be nice to not have to ctrl+o after each time i have invoked the skill

pale2hall · 5 months ago

And on GUI? I keep checking the setting for an "auto expand bash" or similar. I'm a programmer I NEED verbosity!! please

Tectract · 4 months ago

bump please.

How do I actually see the output of my scripts? all I see is "+52 lines" at the bottom. I need to actually see the output from my scripts? Surely there must be a way?

akashagarwal7 · 4 months ago

Bump please.

cubicleDowns · 3 months ago

Double bump please!

cubicleDowns · 3 months ago

The thing is, parsing a skill/bash output burns tokens.... GOOD for Anthropic. Annoying for our workflow.

If the output is JUST text, it's not a threat. If Claude is executing text in a 'cat' that is a Claude problem.

akashagarwal7 · 3 months ago

Has claude code made this worse on purpose?! While reading an expanded output, background commands get stuck on waiting on user input with the user input not visible anymore, until you exit out of it and ask claude to continue. Very, very inconvenient.

Edit - Found out that launching with claude --verbose expands the transcripts always! More from Claude:

Showing cached comments. Read the full discussion on GitHub ↗

[FEATURE] Feature Request: Allow configuration to prevent auto-collapsing of long Bash outputs

Preflight Checklist

Problem Statement

Problem

Proposed Solution

Alternative Solutions

Priority

Feature Category

Use Case Example

Additional Context

14 Comments

Related Claude Code issues

Browse by topic