Analyze pi-agent session usage with the same reporting experience as ccusage.

Quick Start

# Recommended - always include @latest
npx @ccusage/pi@latest --help
bunx @ccusage/pi@latest --help

# Alternative package runners
pnpm dlx @ccusage/pi
pnpx @ccusage/pi

Recommended: Shell Alias

Since npx @ccusage/pi@latest is quite long to type repeatedly, we strongly recommend setting up a shell alias:

# bash/zsh: alias ccusage-pi='bunx @ccusage/pi@latest'
# fish:     alias ccusage-pi 'bunx @ccusage/pi@latest'

# Then simply run:
ccusage-pi daily
ccusage-pi monthly --json

💡 The CLI reads pi-agent session data from ~/.pi/agent/sessions/ (configurable via PI_AGENT_DIR).

Common Commands

# Daily usage grouped by date (default command)
npx @ccusage/pi@latest daily

# Monthly usage grouped by month
npx @ccusage/pi@latest monthly

# Session-based usage
npx @ccusage/pi@latest session

# JSON output for scripting
npx @ccusage/pi@latest daily --json

# Custom pi-agent path
npx @ccusage/pi@latest daily --pi-path /path/to/sessions

# Filter by date range
npx @ccusage/pi@latest daily --since 2025-12-01 --until 2025-12-19

Useful environment variables:

• PI_AGENT_DIR – override the pi-agent sessions directory (defaults to ~/.pi/agent/sessions)
• LOG_LEVEL – control log verbosity (0 silent … 5 trace)

What is pi-agent?

Pi-agent is an alternative Claude coding agent. It stores usage data in a similar JSONL format to Claude Code but in a different directory structure.

Features

• 📊 Daily/Monthly/Session Reports: Same reporting options as ccusage
• 💵 Accurate Cost Calculation: Uses LiteLLM pricing database
• 📄 JSON Output: Export data in structured JSON format with --json
• 📱 Compact Mode: Use --compact flag for narrow terminals

Data Source

Pi-agent session data is read from:

| Directory | Default Path | | ----------------- | ----------------------- | | Pi-agent sessions | ~/.pi/agent/sessions/ |

Documentation

For detailed guides and examples, visit ccusage.com.

Sponsors

Featured Sponsor

Check out ccusage: The Claude Code cost scorecard that went viral

License

MIT © @ryoppippi