VibeControls
Ask AI
Insights2 illustrations

Fleet Health and Live Logs

See which agents, sessions, vibes and targets are healthy, then stream a struggling agent's logs to find out why.

These images are illustrations of the concept, not screenshots of the actual product.

Overview

Fleet health and live logs is the VibeControls concept for answering two questions in quick succession: is anything in the fleet unwell, and what is it doing right now? The health page summarizes the state of every agent, session, vibe and target in the workspace, and the log viewer streams output from a chosen agent in near real time, so spotting a problem and investigating it happen in the same place.

Without a shared view, teams tend to learn that a runner or a training box has gone quiet only when a job fails, and then someone connects over SSH to tail logs by hand. The concept replaces that routine with continuous heartbeats, a status headline that says how many agents are healthy, and a log stream that can be narrowed by severity and text without logging in to the machine.

On the health page, a headline reports the overall state of the system with a one-line summary, and four cards count healthy agents, running sessions, active vibes and connected targets, calling out any degraded agent. An Agent Health list pairs each agent with a heartbeat trace, its last heartbeat and uptime, and a healthy or degraded badge; a degraded agent expands to show its most recent error. The page refreshes on a timer, with an auto-refresh toggle and a manual Refresh button. The Logs page picks an agent and shows a live stream with a timestamp, severity and source on every line, plus a toolbar to change the minimum severity, filter text, refresh, pause, clear and download.

Both screens live under Insights alongside analytics and the audit log. Health is designed to lead straight into the log viewer: the degraded agent flagged on one page is the agent whose stream you open on the other, and its log lines refer to the sessions, vibes, tunnels and coding-agent runs that the rest of VibeControls manages.

What this concept shows

  • System status headline with a one-line summary of agents, sessions, vibes and target connections
  • Summary cards for healthy agents, running sessions, active vibes and connected targets, with degraded counts called out
  • Per-agent heartbeat trace, last heartbeat, uptime and a healthy or degraded badge
  • Expandable error detail on a degraded agent
  • Auto-refresh on a fixed interval with a manual Refresh button
  • Live log stream per agent with a timestamp, severity and source on every line
  • Severity threshold, text filter, entry count and pause, clear and download controls

How it works

  1. Open Health under Insights to read the system status headline and the summary cards.
  2. Scan the Agent Health list for an agent marked degraded and expand it to read its latest error.
  3. Open Logs and select that agent in the agent picker.
  4. Keep the stream live, adjust the minimum severity and type a filter to narrow the lines.
  5. Pause the stream to study a burst of warnings or errors, then download the entries to share them.

Who it's for

  • Platform and DevOps engineers on call for a fleet of agents
  • ML engineers running long training or data pipeline jobs
  • Developers debugging a coding agent that stopped responding
  • Engineering leads who want a quick read on fleet reliability

Illustrations

2 illustrations of this concept. Select one to view it full size.

Fleet Health Overview

Health summarizes agents, sessions, vibes and targets and flags a degraded agent with its latest error.

This illustration shows the Health page under Insights. The header gives the last update time and notes that the page refreshes every 30 seconds, with an Auto-refresh checkbox and a Refresh button. A large status card reports that the system is degraded and summarizes, in one line, how many agents are healthy, sessions running, vibes active and target connections open. Four cards follow for agents, sessions, vibes and targets, each showing a count against the total, with a badge calling out one degraded agent. The Agent Health list gives every agent a row with its name, host details, platform and version, a heartbeat trace, the time since the last heartbeat and uptime, and a healthy or degraded badge with a row menu. The degraded agent's row expands into a red error panel showing its most recent failure. Agent names and counts are sample data.

Live Log Viewer for an Agent

The log viewer streams one agent's output live, with severity and text filters plus pause, clear and download.

This illustration shows the Logs page under Insights, described as log streaming from your agents relayed through the platform. An information banner explains that the logs support troubleshooting, debugging and auditing in near real time. An agent picker selects the machine to watch and shows its operating system and architecture, with a green dot confirming the connection. The viewer toolbar has a Live indicator, a severity threshold set to info and above, a text filter, a count of entries, and buttons to refresh, pause, clear and download. Each line shows a timestamp, a color-coded level of info, debug, warn or error, a source tag of agent, session or tunnel, and a message, such as a terminal session starting for a vibe, a coding-agent run completing, a tunnel connecting or a model load failing for lack of memory. A Select action button sits at the top right. Log lines are sample data.

Topics

  • agent health monitoring
  • fleet health dashboard
  • live log viewer
  • real-time log streaming
  • agent heartbeat monitoring
  • degraded agent alert
  • developer machine monitoring
  • log severity filter
  • tail remote logs in browser
  • coding agent observability

Related concepts

All VibeControls concepts
Get started freeBrowse all concepts

We use cookies for essential site functions and, with your consent, for analytics to improve VibeControls. We don't use advertising or cross-site tracking cookies. See our Cookie Policy.

Preferences