Unofficial. ClaudeDash is a community project and is not affiliated with or endorsed by Anthropic. "Claude" and "Claude Code" are trademarks of Anthropic.

ClaudeDash reads the session logs Claude Code already writes to disk and turns them into a dashboard inside VS Code. It is built for people on a Claude subscription (Pro / Max) who want to know what their usage would have cost at pay-as-you-go API prices — broken down to the individual conversation.

Everything runs locally. No account, no telemetry, no usage data ever leaves your machine. The only optional network request is fetching current model prices.

Features

• Per-conversation cost & tokens — the headline view. Every conversation, titled by its first prompt, with a full drill-down: input, output, cache writes (5-minute and 1-hour tiers), cache reads, and web tools.
• Accurate, cache-aware API-equivalent cost — prices input, output, prompt-cache writes/reads, and server tools separately, and applies the 1M long-context tier automatically when a request's prompt exceeds 200K tokens.
• Configurable status bar — shows cost and input/output tokens for whatever scope you pick: today, this week, this month, all time, the whole current project, or your latest conversation. Switch it in two clicks with ClaudeDash: Choose Status Bar Metric.
• Cost over time — daily spend, stacked by model.
• Breakdowns — by model and by project.
• Budgets — set a monthly cap; the status bar turns amber at your threshold and red when exceeded.
• Filters — by date range, model, and project. Export the filtered view to CSV (per conversation) or JSON.
• Live — a file watcher updates the figures within ~1 second as you work, using an incremental parser so even large histories stay fast.

How cost is calculated

Claude Code writes one JSONL file per session under your projects directory. ClaudeDash reads the usage block of each assistant message and prices each component independently:

Prices come from the LiteLLM pricing dataset by default (fetched once, cached locally, refreshed daily in the background). When you are offline — or if you set claudedash.pricing.source to bundled — ClaudeDash uses a snapshot shipped with the extension. The number is an estimate of equivalent API cost; your actual subscription is a flat fee.

Install

From the Marketplace (once published): search for ClaudeDash in the Extensions view.

From a .vsix:

git clone https://github.com/runtimebug/claudedash.git
cd claudedash
npm install
npm run build
npm run package        # produces claudedash-<version>.vsix
code --install-extension claudedash-*.vsix

Usage

• Run ClaudeDash: Open Dashboard from the Command Palette (or click the status-bar item).
• Click any conversation to see its full breakdown.
• The dashboard opens scoped to your current project — clear the Projects filter (or pick others) to see everything. Range and model filters live alongside it; Export writes the current view.
• ClaudeDash: Choose Status Bar Metric picks what the status bar shows; Pin Conversation for Status Bar locks the "latest conversation" metric to a specific one.
• ClaudeDash: Set Monthly Budget… sets your cap.

Configuration

Privacy

ClaudeDash reads session files locally and never transmits them. The single optional outbound request fetches public model prices from the configured LiteLLM URL; disable it entirely with claudedash.pricing.source: "bundled".

Development

npm install      # installs the host + webview workspaces
npm run build    # bundles the host (esbuild) and webview (Vite)
npm test         # core unit tests (Vitest)
npm run typecheck
npm run lint

Press F5 in VS Code to launch an Extension Development Host. The architecture and contribution workflow are documented in CONTRIBUTING.md.

License

MIT