Sandbox
@a-connoisseur/patch-claude-code

Patched Claude Code binaries with visible tool calls

This repo patches Claude Code’s native binary so the UI shows detailed tool calls, subagent prompts, and live thinking without verbose mode. It also disables spinner tips and renames the header so you can spot the patched build.

46 stars5 forksTypeScriptUpdated 7d ago
Who it's for

Builders who use Claude Code and want clearer agent output while they work.

What it delivers

You can watch what Claude Code is doing in real time instead of waiting on collapsed summaries.

What it does

Detailed tool call display

Shows files read and commands run instead of collapsed tool summaries.

Live thinking stream

Streams Claude’s thinking inline in the UI, even when verbose mode is off.

Subagent prompt visibility

Shows subagent `Prompt:` blocks in the normal interface.

Spinner tip removal

Hard disables spinner tips so the screen stays focused on useful output.

Patched binary identification

Renames the startup header to `Connoisseur's Code v...` so the patch is easy to recognize.

How to get it

  1. 1If you installed Claude Code via npm, remove it and install the native build first
    npm uninstall -g @anthropic-ai/claude-code
    curl -fsSL https://claude.ai/install.sh | bash
    claude --version
  2. 2This installer detects your OS and CPU architecture and downloads the matching patched…
    curl -fsSL https://raw.githubusercontent.com/a-connoisseur/patch-claude-code/main/install-patched-claude.sh | bash
  3. 3On Windows, use PowerShell
    irm https://raw.githubusercontent.com/a-connoisseur/patch-claude-code/main/install-patched-claude.ps1 | iex
  4. 4Run
    chmod +x ./claude.native.patched
    sudo mv ./claude.native.patched "$(which claude)"
    claude --version
  5. 5Run
    chmod +x ./claude.native.macos.patched
    sudo mv ./claude.native.macos.patched "$(which claude)"
    xattr -dr com.apple.quarantine "$(which claude)"
    claude --version
  6. 6Run
    $target = (Get-Command claude).Source
    Copy-Item .\claude.native.windows.patched.exe $target -Force
    claude --version

README

Patch Claude Code

What this does

This repo publishes patched native Claude binaries that make output more transparent without verbose mode. Here is an exhaustive list of things it changes:

  • Shows detailed tool calls instead of collapsed summaries.
  • Hard disables spinner tips.
  • Streams thinking live in the UI. This is helpful for instances where Claude thinks for over 10 minutes and you want to know if it's actually still doing something.
  • Shows subagent Prompt: blocks in the non-verbose UI.
  • Renames the startup header to Connoisseur's Code v... (this makes it easy to identify when Claude has auto updated and lost the patch).

Thinking note:

  • The patched binary requests summarized thinking automatically. You do not need to enable showThinkingSummaries in your Claude settings.

Quick Start

Prerequisite

If you installed Claude Code via npm, remove it and install the native build first:

npm uninstall -g @anthropic-ai/claude-code
curl -fsSL https://claude.ai/install.sh | bash
claude --version

Automatic Install

This installer detects your OS and CPU architecture and downloads the matching patched release for that version and platform.

curl -fsSL https://raw.githubusercontent.com/a-connoisseur/patch-claude-code/main/install-patched-claude.sh | bash

On Windows, use PowerShell:

irm https://raw.githubusercontent.com/a-connoisseur/patch-claude-code/main/install-patched-claude.ps1 | iex

If you'd rather avoid blindly running scripts from the internet, you can do it the manual way below. That said, the binaries are built on Github Actions and the patcher is also free for you to see and modify, so there's no reason to trust this repo or the release builds other than convenience.

Manual Install (From Releases, native only)

  1. Pick the release tag for your platform:

    • macOS arm64: macos-arm64
    • Linux x64: linux-x64
    • Linux arm64: linux-arm64
    • Windows x64: win32-x64
    • Windows arm64: win32-arm64
  2. In that release, download the regular patched binary for your platform.

  3. Follow the install instructions for your platform below.

Install (Linux)

chmod +x ./claude.native.patched
sudo mv ./claude.native.patched "$(which claude)"
claude --version

Install (macOS)

chmod +x ./claude.native.macos.patched
sudo mv ./claude.native.macos.patched "$(which claude)"
xattr -dr com.apple.quarantine "$(which claude)"
claude --version

Install (Windows)

$target = (Get-Command claude).Source
Copy-Item .\claude.native.windows.patched.exe $target -Force
claude --version

Files in the repo

Repository payload11 top-level entries
  • .github
  • scripts
  • test
  • .gitignore
  • bun.lock
  • install-patched-claude.ps1
  • install-patched-claude.sh
  • package.json
  • patch-claude-display.ts
  • PATCHING_PLAYBOOK.md
  • README.md

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More tools

JuliusBrussee/
caveman

🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman

105k
1 add
MemPalace/
mempalace

The best-benchmarked open-source AI memory system. And it's free.

59k
stablyai/
orca

Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.

66k

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

132k

Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors

64k
headroomlabs-ai/
headroom

Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

71k