clawview 0.2.5

Real-time dashboard for monitoring Claude Code sessions

Real-time dashboard for monitoring Claude Code sessions

Features · Quick Start · Development · How It Works

---

ClawView reads the JSONL session files that Claude Code writes to ~/.claude/projects/ and presents them in a live-updating web dashboard. Track token usage, costs, tool invocations, errors, and more across all your coding sessions.

Features

• Live session monitoring -- active sessions update in real time via WebSocket
• Project-level analytics -- token usage over time, daily costs, model breakdown, tool usage trends
• Session deep-dive -- conversation timeline with turn-by-turn token counts, tool calls, and context window usage
• Cost tracking -- per-session and per-project cost estimates based on model pricing
• Tool & MCP tracking -- see which tools and MCP servers each session uses, with category breakdowns
• Error & interruption rates -- spot problematic sessions at a glance
• Memory & skill browser -- inspect memory files and sub-agent skills from within the dashboard
• Continuation chain linking -- follows /clear continuations across sessions
• IDE integration -- links to open files directly in your editor

Screenshots

Quick Start

uvx clawview

That's it. Open http://localhost:3333 in your browser.

Requires uv (install: curl -LsSf https://astral.sh/uv/install.sh | sh)

From source

git clone https://github.com/tuongaz/clawview.git
cd clawview
make run

This requires Python 3.11+, uv, and Bun.

Development

Run the frontend and backend separately for hot-reload:

# Terminal 1 -- Frontend (Vite dev server with HMR)
cd frontend && bun run dev

# Terminal 2 -- Backend
uv run clawview

Other commands

make build          # Build frontend + sync Python deps
make clean          # Remove web/dist, frontend/node_modules, .venv
uv run pytest       # Run tests
uv run pyright src/clawview/  # Type checking

How It Works

ClawView is a Python (FastAPI) backend that serves a React (Vite) frontend as static files.

flowchart LR
    A["~/.claude/projects/**/*.jsonl"] --> B["FastAPI Backend"]
    B -- "WebSocket" --> C["React App (Browser)"]
    C -- "REST / WS" --> B

The backend watches Claude Code's session files, parses JSONL entries into structured data (sessions, turns, tool events), computes analytics, and pushes updates to connected clients over multiple WebSocket channels.

Tech stack

Layer	Technology
Backend	Python, FastAPI, Uvicorn, WebSockets
Frontend	React 19, TypeScript, Vite, Tailwind CSS, HeroUI
Charts	Recharts
Package	uv (Python), Bun (JS)

Credits

Inspired by Sniffly by Chip Huyen.

License

MIT