SuperTrace - Monitor your AI coding sessions

See what your AI assistant is doing. Track inputs, outputs, and tool calls in real-time.

Install | Features | Context Tracking | Configuration | Docker | Troubleshooting

---

---

Install

curl -fsSL https://quickcall.dev/supertrace/install.sh | bash

Then run:

quickcall-supertrace

Open http://localhost:7845 in your browser.

SuperTrace reads directly from Claude Code's JSONL transcript files at ~/.claude/projects/. No hooks or configuration needed.

100% Local - All data stays on your machine. Nothing is sent to any external servers.

Features

• Real-time monitoring - Watch AI assistant inputs/outputs as they happen
• Session timeline - Browse all your coding sessions
• Conversation view - See user prompts, assistant responses, and tool calls
• Full-text search - Find anything across all sessions
• Export - Download sessions as JSON or Markdown
• WebSocket updates - Live updates without page refresh
• Context window tracking - Real-time context usage with color-coded progress bar

Context Window Tracking

Real-time context window tracking is automatically enabled when you run SuperTrace.

How It Works

1. When quickcall-supertrace starts, it automatically configures Claude Code hooks
2. After each Claude response, the hook captures token usage
3. Context data is sent to the SuperTrace server
4. The UI displays a real-time progress bar:
   • Green - Under 50% usage
   • Yellow - 50-75% usage
   • Red - Over 75% usage

Setup

Just run SuperTrace - hooks are configured automatically:

quickcall-supertrace

Then restart Claude Code to load the hooks.

Disable Auto-Registration

If you don't want automatic hook registration:

QUICKCALL_SUPERTRACE_AUTO_HOOKS=false quickcall-supertrace

Environment Variables

Variable	Default	Description
QUICKCALL_SUPERTRACE_AUTO_HOOKS	true	Auto-register Claude Code hooks
QUICKCALL_SUPERTRACE_DEBUG	false	Enable debug logging for hooks

Dashboard Metrics

Hero Metrics (6-Panel Grid)

Metric	Description
Commits	Git commits made during the session
Turns / Commit	Average prompts per commit (lower = faster delivery)
Tool Success Rate	Percentage of tool calls that completed successfully
Lines / Hour	Net lines changed per hour (productivity metric)
Images	Total images/screenshots shared in the session
Thinking	Prompts with extended thinking enabled (e.g., "3/10")

Cost Analysis

Metric	Description
Estimated Cost	Total USD cost based on Claude API pricing
Input Cost	Cost for context/input tokens
Output Cost	Cost for generated tokens
Cache Savings	Money saved from prompt caching

Token Metrics (Per-Turn Chart)

• Input Tokens - Context sent per prompt
• Output Tokens - Tokens generated in response
• Cache Read Tokens - Tokens read from cache
• Cache Creation Tokens - Tokens written to cache
• Turn Duration - Time per prompt/response cycle

Tool Usage

• Tool Distribution - Breakdown by tool type (Read, Edit, Bash, etc.)
• Total Tools - Number of tool calls
• Tools Per Turn - Stacked visualization of tools used

Work Output

Metric	Description
Files Changed	Unique files modified
Lines Added	Lines of code added
Lines Removed	Lines of code removed
Net Lines	Net change (added - removed)
Files Read	Files read for context

AI Insights

• Session Intents - AI-detected goals for the session
• Intent Changes - Whether focus shifted during work

Time Filtering

All metrics support time range filtering: 1h, 2h, 6h, 24h, All

Configuration

Env Variable	Default	Description
QUICKCALL_SUPERTRACE_PORT	7845	Server port
QUICKCALL_SUPERTRACE_HOST	127.0.0.1	Server host

Docker

docker compose up -d

Troubleshooting

Port Already in Use

QUICKCALL_SUPERTRACE_PORT=8080 quickcall-supertrace

Reset Database

rm -rf ~/.quickcall-supertrace

Stop the Server

# Foreground: Ctrl+C# Background: pkill -f quickcall-supertrace

---

Built with care by QuickCall