Real-time monitoring dashboard for your entire AI fleet. Track agent status, token usage, heartbeat, memory, and live output — all in one place.
Three steps to full visibility into your AI agent fleet. No complex setup, no vendor lock-in.
One command to install NerveCenter globally. Works with any Node.js environment — local, cloud, or CI.
→Point NerveCenter to your agents with a simple YAML config. Define names, endpoints, and monitoring intervals.
→Open the dashboard and watch your agents in real time. Status, tokens, heartbeats, memory, and live output — all at a glance.
Purpose-built for AI agent monitoring. Every feature designed around how agents actually work.
See which agents are online, idle, or offline at a glance. Real-time status updates with automatic health detection and alerting when agents go dark.
Monitor token consumption across all agents in real time. Track input/output tokens, cost estimates, and usage trends over time to optimize your spend.
Continuous heartbeat checks ensure you know the moment an agent stops responding. Configurable intervals, grace periods, and escalation rules.
Stream agent output in real time. Watch tool calls, reasoning traces, and responses as they happen — like tailing logs, but for AI agents.
Inspect agent memory and context. See what your agents remember, how context is managed, and debug memory-related issues without guessing.
A unified timeline of everything happening across your agent fleet. Filter by agent, event type, or severity to find exactly what you need.
Full audit trail of every agent session. Review past conversations, tool usage patterns, error rates, and performance metrics over any time range.
A glimpse of the NerveCenter dashboard monitoring a live agent fleet.
Monitoring designed from the ground up for AI agent architectures.
Fully open source under MIT license. Inspect the code, contribute features, fork it — your monitoring, your rules. No black boxes, no vendor lock-in.
Your data stays on your infrastructure. Deploy on a VPS, bare metal, or your local machine. Zero telemetry, zero external dependencies, full control.
Works with any AI framework — LangChain, CrewAI, AutoGen, OpenClaw, or your custom setup. If it speaks HTTP, NerveCenter can monitor it.
Join the community building the future of AI agent monitoring. Star the repo, contribute, or just get started.