ai-sessions is a local-first analytics dashboard for Claude Code session activity. It parses Claude session JSONL files from ~/.claude/projects/, computes usage/cost/productivity metrics, and serves a React dashboard with live updates over WebSocket. What it does Loads historical Claude Code sessions into an in-memory store. Aggregates token, cost, tool, project, and daily activity metrics. Streams near-real-time updates when session files change. Exposes REST APIs for dashboard data and detailed drill-down views. Serves a React + Vite UI from the same Go binary. Who this helps This dashboard is useful for: Individual developers who want to understand AI usage, token burn, and cost over time. Power users of Claude Code who want to improve prompt quality and session hygiene. Tech leads / engineering managers who need visibility into usage patterns, project focus, and productivity windows. Anyone optimizing AI spend who wants to track cache efficiency and reduce repeated context costs. Widget ideas this dashboard gives you If you are building your own analytics UI, this project shows practical widget patterns you can reuse: KPI cards: sessions, total spend, output tokens, cache efficiency. Token trend chart: cache vs input/output views, with date-range toggles. Session explorer table: searchable session list plus detail drawer. Tool usage panel: top tools with clickable sample drill-down. Hourly activity chart: peak coding/agent activity by hour. Conversation breakdown table: prompt/response previews with output, context %, cost, and time. Context health panel: context-window fill tracking with warning states. Prompt insights card: actionable suggestions to reduce token waste and improve quality. Live update banner: real-time feedback when new session data arrives.

AI Usage Dashboard

analytics dashboard for Claude Code session activity.

analytics dashboard for Claude Code session activity.