Install

Homebrew (Recommended)

brew install --cask kangraemin/tap/claude-inspector && sleep 2 && open -a "Claude Inspector"

Direct Download

Download the .dmg from the Releases page.

Upgrade / Uninstall

# Upgrade
brew update && brew upgrade --cask claude-inspector && sleep 2 && open -a "Claude Inspector"

# Uninstall
brew uninstall --cask claude-inspector

Proxy Mode

Intercept real Claude Code CLI traffic via a local MITM proxy.

Claude Code CLI  →  Inspector (localhost:9090)  →  api.anthropic.com

1. Click Start Proxy in the app
2. Run Claude Code through the proxy:

ANTHROPIC_BASE_URL=http://localhost:9090 claude

3. Every API request/response is captured in real-time.

4 Tabs

AI Analysis

Select captured requests and let Claude (Sonnet) analyze the full session flow.

What It Produces

• Step-by-step session summary — what happened, in what order, with mechanism explanations
• MCP dynamic loading chain — how ToolSearch fetches schemas, how deferred tools become callable
• Skill loading flow — how slash commands trigger <command-message> → Skill tool_use
• Sub-agent patterns — which requests are main vs sub-agent, and how they relate
• Mermaid flowchart — visual diagram of the internal mechanism flow
• Clickable references — each step links to the actual Request, scrolling to the relevant tool_use

How to Use

1. Capture some requests via the proxy
2. Click the AI Analysis tab
3. Select requests to analyze (click to toggle, Shift+click for range)
4. Click Analyze Session Flow
5. Watch real-time streaming as Claude processes the session
6. Click any Request #N badge to jump to that request's tool_use in the JSON tree

Inline Chat

After analysis, ask Claude follow-up questions about the session. The chat understands the full analysis context.

Session Detection

Requests are automatically grouped by session using message content fingerprinting. Different sessions (main conversation, sub-agents, stop hooks) get different colored borders in the sidebar.

What You'll Learn

All discovered from real captured traffic. See what Claude Code hides from you.

1. CLAUDE.md is injected into every single request

You type hello. Claude Code silently prepends ~12KB before your message:

Injection order: Global CLAUDE.md → Global rules → Project CLAUDE.md → Memory

This ~12KB payload is re-sent with every request. A 500-line CLAUDE.md quietly burns tokens on every API call. Keep it lean.

2. MCP tools are lazy-loaded — watch tools[] grow

Built-in tools (27) ship their full JSON schemas every request. MCP tools start as names only.

Unused MCP tools never consume tokens. Watch tools[] grow as the model discovers what it needs.

3. Images are base64-encoded inline

When Claude Code reads a screenshot or image, it's base64-encoded and embedded directly in the JSON body:

{
  "type": "image",
  "source": {
    "type": "base64",
    "media_type": "image/png",
    "data": "iVBORw0KGgo..."
  }
}

A single screenshot can add hundreds of KB to the request payload. Inspector shows you the exact size.

4. Skill ≠ Command — completely different injection paths

Typing /something triggers one of three completely different mechanisms:

Commands run locally and only pass the result. Skills inject the entire prompt text — and it stays in every subsequent request until the session ends.

5. Previous messages pile up — use /clear often

Claude Code re-sends the entire messages[] array with every request:

Most of it is old conversation you no longer need. Running /clear resets the context and drops the accumulated weight.

6. Sub-agents run in fully isolated contexts

When Claude Code spawns a sub-agent (via the Agent tool), it creates a completely separate API call:

Inspector captures both calls side by side, and AI Analysis automatically detects and labels sub-agent sessions.

Development

git clone https://github.com/kangraemin/claude-inspector.git
cd claude-inspector
npm install
npm start          # Dev mode
npm run test:unit  # Unit tests
npm run test:e2e   # E2E tests (Playwright)

Tech Stack

Privacy: All proxy traffic stays on localhost. AI Analysis runs locally via claude -p (your own Claude Code CLI).

License

MIT