claude-code-hooks-multi-agent-observability
Real-time monitoring for Claude Code agents through simple hook event tracking.
Claude Code Hooks Multi-Agent Observability is a comprehensive monitoring system for Claude Code agents that captures, stores, and visualizes hook events in real-time. It intercepts all 12 Claude Code lifecycle events—from tool execution and subagent orchestration to context compaction and session management—sending them to a dashboard for live inspection. The system uses Python hook scripts, a Bun-powered server with SQLite storage, and a Vue 3 client with WebSocket streaming to provide complete visibility into multi-agent workflows. Developers copy the `.claude` directory into their projects to enable event tracking across concurrent agent teams.
Key Features
Use Cases
- 01Monitoring multiple concurrent Claude Code agents working in parallel on complex engineering tasks
- 02Debugging tool execution failures and permission issues across subagent workflows
- 03Tracking context window usage and compaction events during long-running agent sessions
- 04Visualizing multi-agent orchestration patterns with task handoffs and inter-agent messaging
- 05Auditing tool usage patterns and session behavior across different projects and repositories
- 06Validating agent team plans and ensuring required specification sections are completed before task execution
Related Plugins
View moreandrej-karpathy-skills
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
claude-mem
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
Understand-Anything
Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
rtk
CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies
claude-code-hooks-multi-agent-observability — FAQ
What is claude-code-hooks-multi-agent-observability?+
It's a real-time monitoring and visualization system for Claude Code agents that captures all hook events (tool usage, subagent lifecycle, notifications, etc.) and displays them in a live dashboard. It works by intercepting Claude Code's 12 lifecycle hooks through Python scripts that send event data to a server and client for visualization.
How do I install and set up the observability system?+
First, install prerequisites: Claude Code CLI, Astral uv, Bun/npm/yarn, and optionally just. Clone the repo, run `just start` (or `./scripts/start-system.sh`) to launch the server and client, then copy the `.claude` directory to any project you want to monitor. Update `settings.json` in each project to set your `--source-app` identifier, and events will stream to http://localhost:5173.
Which clients does this work with?+
This system is specifically designed for Claude Code, Anthropic's official CLI agent. It hooks into Claude Code's native hook system and requires Claude Code to be installed and configured with an Anthropic API key.
What API keys or prerequisites are required?+
Required: Anthropic API key (for Claude Code) and Astral uv (Python package manager). Optional: OpenAI API key (multi-model support), ElevenLabs API key (TTS features), and Firecrawl API key (web scraping). You'll also need Bun, npm, or yarn to run the server and client.
Is this observability system free and open source?+
The observability system code itself is open source on GitHub. However, you'll need to pay for API usage from Anthropic (Claude Code requires credits), and optionally OpenAI, ElevenLabs, or Firecrawl if you use those integrations.
Can I monitor multiple projects and agent teams simultaneously?+
Yes, you can copy the `.claude` directory to multiple projects, each with a unique `--source-app` identifier. The dashboard filters events by application and session, allowing you to track concurrent agents across different codebases and monitor multi-agent orchestration with parallel subagent teams.
How do I install claude-code-hooks-multi-agent-observability?+
Open the source repository on GitHub and follow its README. claude-code-hooks-multi-agent-observability is a plugin — MCP Agents Market links you directly to the official repo.
Is claude-code-hooks-multi-agent-observability free?+
claude-code-hooks-multi-agent-observability is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.