Holus Observatory - Multi-Agent Monitoring Dashboard
Holus Observatory is the monitoring frontend for the Holus multi-agent system. It reads from the agent registry, trajectory logs, evaluation history, and content queue to provide a real-time view of what agents are doing, how well they perform, and what content they produce.
The Problem
When AI agents run autonomously, you need visibility into what they are doing, which ones are performing well, and where quality is dropping. Without observability, autonomous systems become black boxes.
The Solution
A Next.js dashboard that reads from the Holus backend (AGENTS.yaml, trajectory.jsonl, eval_history.jsonl, content-queue) and displays: agent grid with status badges, quality heatmap (agents x days), content pipeline kanban, engagement tracker with platform filters, follower growth charts, and system health monitoring.
The Outcome
Recruiters can inspect a working observability surface, and operators can identify quality drops, stalled agents, or content bottlenecks at a glance.
Key Features
- Agent grid with live status badges (active/idle/running)
- Quality heatmap - agents x days with color-coded scores
- Content pipeline kanban (draft → review → published)
- Engagement tracker with platform filters and sparkline charts
- Follower growth chart with daily net change bars
- System health monitoring with service latency
Technology Stack
Interested in this project?
I'd love to discuss the technical details, challenges overcome, or similar projects I could build for you.
Let's discuss this project