Does this skill require an external database?

No, it uses a lightweight filesystem-based approach with JSONL files and in-memory streaming to ensure high performance and zero database overhead.

Can I track subagents running in parallel?

Yes, the dashboard features 'swim lane' visualizations specifically designed to track and display parallel execution across multiple agents such as designers, engineers, and testers.

What event types are captured by the dashboard?

It captures a comprehensive range of events including session starts, user prompts, pre/post tool execution, and subagent lifecycle events like stops and starts.

How do I access the dashboard interface?

Once the Bun server and client are running locally, you can access the interactive visualization dashboard through your web browser at http://localhost:5173.

Agent Observability Dashboard

Name: Agent Observability Dashboard
Author: rafaelcalleja

byrafaelcalleja

•

分析与监控

Provides real-time visualization and monitoring of multi-agent Claude Code sessions through a high-performance web dashboard.

关于

The Agent Observability skill transforms the way developers monitor Claude Code by providing a visual interface for complex multi-agent workflows. It captures events via filesystem-based hooks and streams them in real-time to a Vue-powered dashboard, allowing users to observe parallel agent execution through intuitive 'swim lane' visualizations. This tool is essential for debugging tool usage, monitoring subagent tasks, and analyzing session flows without the overhead of a database, making it a critical asset for building and optimizing complex AI-driven automation systems.

主要功能

Live analytics charts for monitoring tool usage patterns
Real-time event streaming via WebSockets and Bun server
1 GitHub stars
Multi-agent swim lane visualization for parallel task tracking
Detailed event filtering by agent, session, and event type
Lightweight filesystem-based storage using JSONL format

使用场景

Monitoring long-running Claude Code sessions in real-time for progress tracking
Debugging complex multi-agent interactions and subagent task handoffs
Auditing tool execution and output history for workflow performance optimization

关于

主要功能

Live analytics charts for monitoring tool usage patterns
Real-time event streaming via WebSockets and Bun server
1 GitHub stars
Multi-agent swim lane visualization for parallel task tracking
Detailed event filtering by agent, session, and event type
Lightweight filesystem-based storage using JSONL format

使用场景

Monitoring long-running Claude Code sessions in real-time for progress tracking
Debugging complex multi-agent interactions and subagent task handoffs
Auditing tool execution and output history for workflow performance optimization