ClaudeSavvy Blog

We've shipped a significant update to ClaudeSavvy's dashboard — making it faster to go from a high-level overview to the exact detail you need, and adding three new dedicated views: Conversations, Sub-Agents, and Teams.

Clickable Dashboard Metrics

Every summary card on the dashboard is now a direct link. See an unusually high cache miss rate? Click it — you land straight on the Tokens page. Curious about your top conversation cost? Click it and you're on the Conversations page. No more hunting through the nav.

The same pattern applies to the new Explore Your Usage section at the bottom of the dashboard — six cards giving you a snapshot of Tools & Features, Sub-Agents, Files, Integrations, Conversation Analytics, and Projects, each with a one-click link to dig deeper.

Conversation Analytics

The new Conversations page gives you a complete view of every conversation Claude Code has had across your projects:

• Per-session cost: See exactly what each conversation cost, colour-coded by spend
• Turn count: Understand how long each session ran
• Context window usage: A visual bar shows how much of the context window was consumed — great for spotting conversations that are getting bloated
• Token speed: Avg tokens/coin to understand efficiency

Conversations are sortable and filterable by time period. The dashboard also shows your three most recent sessions at a glance, linking directly to the full list.

Sub-Agents View

Sub-agent usage is now tracked with much more granularity. The Sub-Agents page shows:

• Total exchanges, tokens, cost, and average tokens per exchange
• A timeline chart of sub-agent activity over time
• A detailed table of recent exchanges — including the agent's description, tools called, duration, and cost per invocation

This is particularly useful when you're using Claude Code's Agent tool heavily — you can see exactly which agent tasks are expensive and how long they took.

Teams View

The Teams page tracks token usage and costs across Claude Code's Teams feature. Summary cards show total teams, tokens consumed, and overall cost. When the Teams feature is actively used, a detailed breakdown by team will appear here automatically.

Recent Conversations on the Dashboard

The dashboard now includes a Recent Conversations table showing your last three sessions with project, timestamp, turn count, cost, and context usage — so you get a quick read on activity without leaving the home screen.

How to Get the Update

pip install --upgrade claudesavvy

Then restart with claudesavvy and explore the new Conversations and Sub-Agents pages from the navigation bar.

As always, questions and feedback welcome on GitHub!

We're excited to introduce Teams Analytics in ClaudeSavvy v2.5.0—comprehensive tracking for the Claude Code Teams feature that shows you exactly how much each team costs and which projects are using teams.

Why Teams Analytics Matters

Claude Code's Teams feature lets you spawn multiple agents to work in parallel on complex tasks. While powerful, it can be hard to track:

• How much does each team cost?
• Which teams consume the most tokens?
• Are teams being used efficiently across projects?
• What's the breakdown of input vs output tokens per team?

Now you can answer all these questions with the new Teams page.

What You'll See

The Teams page provides a complete breakdown of your team usage:

• Team Summary Cards: Total teams, tokens consumed, and overall cost at a glance
• Token Distribution Chart: Visual bar chart showing how each team uses Input, Output, Cache Read, and Cache Write tokens
• Detailed Usage Table: Messages, sessions, projects, and complete token breakdown per team
• Time Filtering: Filter by 15min, 1hour, Today, Week, Month, or All time

How It Works

Claude Code automatically logs team activity with a teamName field whenever the Teams feature is invoked. ClaudeSavvy reads this data from your session files and aggregates it into meaningful analytics.

Each team entry shows:

• Message Count: How many messages were processed by the team
• Sessions: Number of unique sessions the team was active in
• Projects: How many different projects used this team
• Token Breakdown: Input, Output, Cache Read, and Cache Write tokens
• Estimated Cost: Calculated using your configured pricing

Accessing Teams Analytics

To view your team usage:

1. Upgrade to ClaudeSavvy v2.5.0: pip install --upgrade claudesavvy
2. Start the dashboard: claudesavvy
3. Click the Teams link in the navigation menu
4. Use the time filters to explore different periods

Under the Hood

This feature was built by enhancing our session parser to extract the teamName field from Claude Code's session logs. When teams are spawned, each message is tagged with the team name, allowing us to aggregate usage statistics per team.

The implementation includes:

• Extended SessionMessage dataclass with team_name field
• New get_team_stats() method for aggregating team usage
• Chart visualization for token distribution across teams
• HTMX-powered time filtering for instant updates

What's Next

We're exploring additional team analytics features:

• Individual teammate/agent tracking within teams
• Team efficiency comparisons
• Cost allocation reports per project/team combination

Questions or feedback? Let us know on GitHub!

We've just shipped a major upgrade to ClaudeSavvy that gives you unprecedented visibility into how Claude Code works. The Sub-Agents and Features pages now offer detailed, interactive analytics that let you explore every tool call and sub-agent invocation.

The Problem We Solved

Previously, you could see aggregate stats—how many times a tool was used, total tokens consumed. But you couldn't answer questions like:

• What files did Claude read during that conversation?
• What commands did it run?
• How do tool calls relate to each other over time?
• Which sub-agent task consumed the most tokens?

Now you can.

Unified Tools Timeline

The Features page now includes a Unified Timeline tab that shows all tool calls across all tools in a single interactive chart. Each tool type has its own color:

• ● Read - File reads
• ● Write - File writes
• ● Edit - File edits
• ● Bash - Shell commands
• ● Glob/Grep - Search operations
• ● Task - Sub-agent launches

Click any bubble to see the exact parameters—which file was read, what command was executed, what search pattern was used.

Zoomable Date Range

When you have hundreds of tool calls, the timeline can get crowded. That's why we added a date range slider below the chart. Drag the handles to zoom into any time period and the chart updates instantly—no page reload required.

Clickable Tool Details

On both the Features and Sub-Agents pages, clicking any row now opens a detailed panel showing:

• Timeline view with all invocations plotted over time
• Conversations breakdown showing usage per session
• Token breakdown with input, output, cache read, and cache write splits
• Individual invocation details when you click a data point

Filter by Session

Want to see what happened in a specific conversation? Use the session filter dropdown to focus on a single session. This is incredibly useful for debugging or understanding how Claude approached a particular task.

What's Next

This is just the beginning of our deep-dive analytics work. We're planning to extend this level of detail to other pages in ClaudeSavvy—watch this space for updates on conversation-level analysis, model comparison views, and more.

Try It Now

Upgrade to the latest version:

pip install --upgrade claudesavvy

Then head to the Features page and click the Unified Timeline tab to explore your tool usage like never before.

Questions or ideas for what we should add next? Let us know on GitHub!

We're thrilled to introduce Efficiency Trend Analysis in ClaudeSavvy v2.2—a smarter way to understand whether your Claude Code usage is actually improving over time.

The Problem with Raw Numbers

Here's something that trips up a lot of developers: if your costs dropped 20% this week, is that actually good? Not necessarily. If your usage also dropped 20%, you haven't improved efficiency at all—you just used Claude less.

That's why we built trend analysis with normalized metrics. Instead of showing raw totals, we compare efficiency metrics that account for activity volume.

What We Track

When you select a time period (Week, Month, or Today), ClaudeSavvy now compares your current metrics against the equivalent previous period:

• Cost per Session: Are you spending more or less per conversation?
• Cost per Message: How efficient is each interaction?
• Tokens per Message: Is your context window staying clean or getting bloated?
• Cache Hit Rate: Are you effectively reusing cached prompts?

Color-Coded Insights

Each metric shows a clear trend indicator:

• ↓ Green = Improvement (costs down, cache hits up)
• ↑ Red = Needs attention (costs up, efficiency down)
• — Gray = No significant change

Context Window Hygiene

The new Tokens per Message metric is particularly useful. If this number keeps climbing, it means your conversations are accumulating context—old messages, file contents, and history that inflate every subsequent API call. A rising trend here is a signal to:

• Start fresh conversations more frequently
• Use /compact to summarize long sessions
• Avoid including unnecessary context in prompts

Tool Cost Trends on Features Page

We've also added trend analysis to the Features page. The tool usage table now shows:

• Cost per Call: How much does each tool invocation cost?
• Trend: Is this tool getting more or less expensive to use?

This helps you identify which tools are driving costs and whether their efficiency is improving.

Smart Recommendations

When trends indicate a problem, ClaudeSavvy generates actionable recommendations. For example:

• Cost per session up 25%? We'll suggest reviewing model selection and conversation length
• Tokens per message up 30%? We'll recommend context window cleanup strategies
• Cache hit rate down 10%? We'll explain how stable prompts improve caching

How to Access

Trend analysis appears automatically when you:

1. Go to the Projects page
2. Select a bounded time period (Today, Week, or Month)
3. Click Analyze on any project

The "Efficiency Trends" section will show your normalized metrics compared to the previous equivalent period.

Upgrade Now

Get the latest version with trend analysis:

pip install --upgrade claudesavvy

We believe understanding real efficiency changes—not just raw numbers—is key to optimizing your Claude Code workflow. Give it a try and let us know what you think!

Questions or feedback? Open an issue on GitHub. Happy tracking! 🚀

We're excited to announce a powerful new feature in ClaudeSavvy: custom rate settings for each LLM model you use with Claude Code!

Why Custom Rates Matter

Every development environment is unique. Whether you're running local models, using enterprise pricing, or accessing models through different providers, having accurate cost tracking is essential. That's why we built the new Settings page to give you full control over your rate configuration.

Key Features

• Model-Specific Pricing: Set custom input/output rates for each LLM model you use
• Real-Time Updates: Changes take effect immediately across all your analytics
• Flexible Configuration: Support for standard pricing, local model rates, or enterprise tiers
• Accurate Cost Tracking: Get precise cost calculations based on your actual pricing

How It Works

The new Settings page provides a clean, intuitive interface where you can:

• View all configured models in one place
• Set custom input token rates (per million tokens)
• Set custom output token rates (per million tokens)
• Easily switch between different pricing tiers
• Save your configuration with a single click

Perfect For:

• 🖥️ Local Models: Track costs for models running on your own hardware
• 🏢 Enterprise Plans: Use your custom enterprise pricing tiers
• 🔄 Multiple Providers: Set different rates for each API provider
• 📊 Budget Planning: Accurately forecast costs based on your actual rates

Getting Started

To access the new Settings page:

1. Make sure ClaudeSavvy is running (claudesavvy)
2. Navigate to the Settings tab in the dashboard
3. Enter your custom rates for each model
4. Click "Save Settings" to apply your changes

Your cost tracking will immediately reflect the new rates, giving you accurate insights tailored to your specific setup.

What's Next?

We're continuing to improve ClaudeSavvy based on your feedback. Future enhancements will include:

• Import/export settings configurations
• Preset templates for common providers
• Historical rate tracking and comparisons

Have questions or suggestions? We'd love to hear from you! Open an issue on GitHub or join the discussion.

Happy tracking, and here's to more accurate cost insights! 🎯

We're excited to announce that Monitor Claude Code is getting a fresh new name: ClaudeSavvy!

Why the Change?

As our tool has evolved, we wanted a name that better reflects its mission—helping you become savvy about your Claude Code usage. The new name is:

• More Memorable: ClaudeSavvy is catchy and easy to remember
• Clearer Purpose: "Savvy" implies smart, intelligent insights into your usage
• Better Branding: Shorter, punchier, and stands out in the Claude Code ecosystem

What's Changing?

Don't worry—the tool you love is staying the same! Only the name is changing. Here's what you need to know:

New Installation Command

Going forward, install ClaudeSavvy using the new package name:

pip install claudesavvy

New Run Command

Start the dashboard with the new command:

claudesavvy

Migrating from Monitor Claude Code

If you're already using the old monitor-claude package:

• Uninstall the old package: pip uninstall monitor-claude
• Install the new package: pip install claudesavvy
• Your existing data and configuration will be preserved
• All your metrics and history will continue to work

What's Staying the Same

• ✅ All the features you know and love
• ✅ The same powerful real-time dashboard
• ✅ Cost tracking and analytics
• ✅ Project-based insights
• ✅ Cache efficiency monitoring
• ✅ 100% free and open-source

What's Next?

We're committed to making ClaudeSavvy the best Claude Code monitoring tool available. The rebrand opens up exciting possibilities for future features and improvements. Stay tuned for more updates!

Thank you for being part of our community. Here's to smarter, more efficient Claude Code usage with ClaudeSavvy! 🚀

Questions? Check out our homepage or open an issue on GitHub.

We're excited to announce the launch of ClaudeSavvy, a free and open-source dashboard for tracking your Claude Code usage, analyzing API consumption, and monitoring costs in real-time.

Why We Built This

As developers increasingly rely on Claude Code for AI-assisted programming, understanding usage patterns becomes essential for managing API costs and optimizing workflows. We needed a tool that could:

• Track Claude Code usage across multiple projects in real-time
• Provide detailed analysis of token consumption and costs
• Help identify optimization opportunities
• Keep data local and private

Key Features

• Real-Time Dashboard: View your API usage metrics as they happen
• Cost Tracking: Monitor every cent spent on Claude API calls
• Project Analytics: Per-project breakdowns of token usage and costs
• Project Analysis: Get direct suggestions for optimizations
• Cache Efficiency: Monitor prompt cache performance and savings
• Tool Usage: See which Claude Code tools you use most

Getting Started

Installation is simple:

pip install claudesavvy
claudesavvy

Then open http://localhost:5000 to view your analytics dashboard.

This is just the beginning! We have many more features planned for future releases. Stay tuned for updates, and feel free to contribute on GitHub!