Claude Code Analytics

Welcome to Claude Code Analytics

Load your ~/.claude directory to view usage statistics.

Click "Load .claude Directory" and select your ~/.claude folder.

Sessions per Day

Session Statistics

Session Details (from session logs)

Token Usage by Model

Cache Efficiency

Cost Breakdown by Model

Cache Savings Analysis

How Prompt Caching Works

What is Prompt Caching?

Claude Code automatically caches your conversation context and system prompts. When you continue a session or start a new one in the same project, Claude can reuse this cached context instead of processing it fresh each time.

1. Cache Write

1.25x base price

First time sending context

→

2. Cache Read

0.1x base price (90% off!)

Reusing cached context

Pricing Breakdown (per Million Tokens)

Token Type	Opus 4	Sonnet 4	Multiplier
Base Input	$15.00	$3.00	1.0x
Output	$75.00	$15.00	—
Cache Write (5min)	$18.75	$3.75	1.25x
Cache Read	$1.50	$0.30	0.1x

Why This Matters

90% Cost Reduction

Cache reads cost just 10% of normal input tokens

Automatic

Claude Code handles caching automatically - no configuration needed

5-Minute TTL

Cache expires after 5 minutes of inactivity (1-hour option available)

Source: Claude API Pricing Documentation

Tool Calls Over Time

Tool Usage by Type

Tool Names & Usage

Tool Usage Insights

Daily Tool Breakdown

📋 Task & Agent Tools

🤖 Task

Launches specialized sub-agents to handle complex, multi-step tasks autonomously. Sub-agents can explore codebases, run tests, plan implementations, or perform research.

Types: Explore Plan Bash general-purpose

📤 TaskOutput

Retrieves output from running or completed background tasks. Used to check on agent progress or get results from async operations.

✅ TodoWrite

Creates and manages structured task lists during coding sessions. Tracks progress, organizes complex tasks, and provides visibility into work items.

❓ AskUserQuestion

Prompts the user with questions during execution to gather preferences, clarify instructions, or get decisions on implementation choices.

🔧 Other Tools

🌐 WebFetch

Fetches and processes content from URLs. Converts HTML to markdown and analyzes web content to extract relevant information.

🔍 WebSearch

Searches the web to find up-to-date information beyond the knowledge cutoff. Returns search results with links and summaries.

📝 ExitPlanMode / EnterPlanMode

Controls planning workflow. EnterPlanMode starts research and design phase; ExitPlanMode signals readiness for user approval.

🛑 KillShell

Terminates running background bash shells. Used to stop long-running processes or clean up stuck commands.

⚡ Skill

Invokes specialized skills/slash commands like /commit, /review-pr for common workflows.

🖥️ Bash Execution Details

Detailed analysis of Bash commands executed by Claude Code (from session logs)

📂

Load ~/.claude directory to see detailed Bash execution data.

Activity by Project

Project Breakdown

Daily API Requests (Messages)

API Calls by Type

API Request Statistics

Data Transfer Estimates

Custom Metric Builder

Metric Name

Formula

Available Variables

totalMessages - Total message count

totalSessions - Total session count

totalToolCalls - Total tool invocations

activeDays - Days with activity

inputTokens - Total input tokens

outputTokens - Total output tokens

cacheReadTokens - Cache read tokens

totalCost - Estimated total cost ($)

Example Metrics to Try

Productivity Score

totalMessages / activeDays

Messages per active day - measures daily output

Tool Efficiency %

(totalToolCalls / totalMessages) * 100

Percentage of messages that involve tool calls

Cost per Session

totalCost / totalSessions

Average cost for each session

Cache Efficiency %

(cacheReadTokens / (cacheReadTokens + inputTokens)) * 100

Percentage of input served from cache

Click any example to load it into the builder

Your Custom Metrics

Model Distribution

Turn Type Distribution

Conversation History