skills.rest
multicam
multicamCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

agent-observability

Stream Claude Code session events to a real-time WebSocket dashboard.

Updated Aug 27, 2026
Search Tags:#Claude Code#multi-agent#observability#workflow visualization#real-time dashboard#AI monitoring#agent debugging
One-click install
npx skills add https://github.com/multicam/qara --skill agent-observability-multicam
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-observability
Source: https://github.com/multicam/qara/tree/main/.claude/skills/agent-observability
Command: npx skills add https://github.com/multicam/qara --skill agent-observability-multicam

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires bun, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides a real-time, visual dashboard to monitor your AI agents' activities, making complex multi-agent interactions transparent and easy to understand. It eliminates the guesswork in debugging and optimizing AI workflows, allowing you to observe exactly how your agents are performing.

Core Features & Use Cases

  • Real-time Event Streaming: Watch agent interactions, tool usage, and session flows as they happen via a WebSocket-powered web dashboard.
  • Agent Swim Lanes: Visualize multiple agents running in parallel, understanding their concurrent tasks and dependencies at a glance.
  • Event Filtering & Search: Easily filter events by agent, session, or type, and search payloads to quickly pinpoint specific interactions or issues.
  • Use Case: When running a complex multi-agent research task, use this dashboard to see which agents are active, which tools they're using, and identify any bottlenecks or errors in real-time.

Quick Start

Start the agent-observability dashboard to monitor my current Claude Code session.

Frequently Asked Questions about agent-observability

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I monitor multiple AI agents running in parallel with real-time visualization?▼

Real-time agent monitoring streams live event data through a WebSocket dashboard showing concurrent agent activities, tool usage, and swim-lane visualizations. Set up Claude Code hooks, configure PAI_DIR, and run the agent-observability dashboard to watch multiple agents execute tasks simultaneously with filterable views by agent, session, or event type.

What's the best way to debug multi-agent workflows and identify bottlenecks?▼

Agent workflow debugging uses real-time event streaming and swim-lane visualization to expose exact interaction sequences, tool calls, and performance issues across parallel agents. Filter events by agent or session type to pinpoint errors, track dependencies, and observe which agents are blocked or competing for resources.

Can I use agent observability with Claude Code sessions running in Bun?▼

Agent observability works with Claude Code environments using Bun runtime, requiring PAI_DIR configuration and Claude Code hooks enabled. The dashboard collects session event data via WebSocket and buffers events in memory with JSONL-based filesystem history for filtering and visualization support.

How do I search and filter agent events to find specific interactions or errors?▼

Event filtering enables searching agent interactions by agent name, session ID, or event type through the dashboard interface. Payload search capabilities let you quickly locate specific tool calls, errors, or data exchanges within multi-agent sessions without parsing raw logs.

What happens to agent event data when the observability dashboard runs?▼

Event data streams live via WebSocket to the dashboard for real-time visualization and is buffered in memory for active sessions. Historical events persist as JSONL files in the filesystem, enabling replay, filtering, and swim-lane reconstruction across past and present agent executions.

Related Skills

Explore More Software Engineering Skills→
rafaelcallejarafaelcalleja
2

agent-observability

Stream Claude Code agent events to a filesystem-based observability dashboard.

Community
Advanced
curiositechcuriositech
10

dag-performance-profiler

Profile DAG execution performance across latency, tokens, cost, and resource usage.

Official
Advanced
zmrezmre
1

agent-observability

Visualize Claude Code agent interactions from JSONL logs via WebSocket streaming.

Community
Advanced
bobmatnycbobmatnyc
68

datadog-observability

Integrate Datadog APM, logs, metrics, synthetics, and RUM for full-stack observability.

Community
Advanced