This project provides a real-time dashboard for monitoring Claude Code sessions across multiple projects. It targets users managing complex AI development workflows, offering enhanced visibility into session status, AI-generated insights, and associated development pipeline information to improve productivity.

How It Works

The system features a daemon and a UI. The daemon watches local Claude Code session logs (~/.claude/projects/), parses them, and uses an XState state machine to derive session statuses (Working, Needs Approval, Waiting, Idle). It generates AI summaries via the Claude Sonnet API and tracks Git PR/CI statuses. Updates are published in real-time using Durable Streams. The React-based UI subscribes to these streams, displaying sessions on a Kanban board organized by repository, showing goals, summaries, and PR/CI details. This architecture ensures immediate feedback and a centralized, dynamic view of development progress.

Quick Start & Requirements

Install dependencies with pnpm install. For accurate "Needs Approval" detection, set up the PermissionRequest hook via pnpm run setup. The daemon requires an Anthropic API key (export ANTHROPIC_API_KEY=sk-ant-...). Start the application with pnpm start (daemon and UI) or separately with pnpm serve (daemon) and pnpm dev (UI). Key dependencies include @durable-streams/* and xstate.

Highlighted Details

• Real-time session updates via Durable Streams.
• Kanban board visualization by session status (Working, Needs Approval, Waiting, Idle).
• AI-powered summaries using Claude Sonnet.
• Integrated tracking of GitHub PR and CI statuses.
• Multi-repository support, grouped by project.

Maintenance & Community

No specific details regarding contributors, sponsorships, or community channels were found in the provided README content.

Licensing & Compatibility

The README content does not specify the project's license type or provide compatibility notes for commercial use or integration with closed-source projects.

Limitations & Caveats

Accurate "waiting_for_approval" status detection relies on the PermissionRequest hook; otherwise, the daemon uses less precise heuristics. AI summary functionality requires a valid Anthropic API key.