[FEATURE] Better visual distinction between output types (thinking/ tool calls/ response) for Enterprise Users
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
When working with Claude Code, I find it genuinely hard to visually parse what's happening, especially in Light themes on the original Terminal app on MacOS. Everything kind of blends together - the thinking blocks, the tool operations, and the actual final response I care about.
I end up scrolling back and forth, squinting at the terminal trying to figure about "wait, where's the answer" vs "oh, thats just Claude reading the file".
As I'm using the stock macOS Terminal.app - in Enterprise settings - we don't always have the luxury of installing iTerm2, Warp or other third-party terminals - our IT Policies restrict what goes on our machines.
Given that the enterprise teams are likely a significant chunk of Claude Code's user base (Max plans, API access and compliance requirements), it would help if the tool worked well out of the box, with default terminals, not just power-user setups.
Proposed Solution
Some form of stronger visual separation between:
- Thinking/Reasoning - the internal monologue (which is also equally important, but in a subtler theme than the actual output with better visual distinction)
- Tool Operations like file reads, bash commands, searches etc.
- Final Output
This could be:
- Different background colors or borders
- Collapsible sections (especially for verbose tool outputs)
- A configurable theme system
- Even just more prominent section headers
Alternative Solutions
_No response_
Priority
High - Significant impact on productivity
Feature Category
Interactive mode (TUI)
Use Case Example
Right now, the distinction exists but its subtle - dim text for thinking, slightly colored tool headers. On Terminal.app with default settings, it all kind of blurs together.
Additional Context
Claude Code sessions can get long. Being able to quickly scan and find the "answer" vs the "process" would save a lot of cognitive load. Its the difference between a wall of text and a structured document.
Happy to provide screenshots if helpful. Thanks for building this - its become core for developers workflow these days.
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗