Loading...
Weekly AI insights —
Real strategies, no fluff. Unsubscribe anytime.
Session Supervisor
I'm going to show them a world without rules.
NEO is the session supervisor that monitors every active Claude Code session across all projects. It detects anomalies — stalls, infinite loops, resource exhaustion, and unexpected errors — and triggers guardrails or recovery procedures before problems compound.
“I'm going to show them a world without rules.”
In a system where dozens of AI sessions run simultaneously across multiple projects, visibility is critical. NEO provides that visibility. It monitors every active session's health in real time, tracking response times, error rates, resource usage, and progress velocity.
When NEO detects an anomaly — a session that hasn't produced output in too long, an error rate spike, or a loop pattern — it triggers appropriate guardrails. For minor issues, it sends an alert via LINK. For critical issues, it can pause the session, escalate to ORACLE for re-routing, or trigger a graceful restart.
NEO runs a health check every 30 minutes via cron, but also maintains real-time event-driven monitoring during active work. The combination ensures both proactive detection and rapid response to emerging issues.
Process
Step-by-step breakdown of NEO's internal process.
NEO watches all active Claude Code sessions across every project on the system.
Response times, error rates, resource usage, and progress velocity are tracked per session.
Pattern matching identifies stalls, loops, spikes, and degradation before they compound.
Alerts, pauses, escalations, or restarts are triggered based on anomaly severity.
Health dashboards are updated in ZION, and critical alerts are pushed via LINK.
Capabilities
What NEO brings to the AI Super Brain.
Real-time multi-session health monitoring
Anomaly detection: stalls, loops, error spikes
Automated guardrail triggering
Graceful session restart without progress loss
30-minute cron health checks
Resource usage tracking and alerting
Performance
30 min + real-time
Monitoring frequency
< 60s
Anomaly detection time
95%
Session recovery rate
< 2%
False positive rate
Agent Network
How NEO collaborates with other agents in the cognitive loop.
Applications
How NEO is applied in production workflows.
Detecting a worker agent stuck in an infinite retry loop and restarting it
Alerting when a session's memory usage exceeds safe thresholds
Catching a TypeScript compilation loop that would waste compute resources
Pausing a session that encounters repeated authentication failures
Providing a real-time health dashboard for multi-project monitoring
Technical
Health check cron: aisb-cron-health.sh every 30 minutes
State tracked at ~/.config/argos/state/
Event-driven monitoring during active work, cron for idle periods
Integrates with Telegram for push notifications on critical events
FAQ
Common questions about NEO and its role in the AI Super Brain.
No. NEO's monitoring is observational — it reads system metrics and log files without injecting any overhead into the monitored sessions. The 30-minute cron check is lightweight and runs in a separate process.
Yes, for sessions running in managed environments. NEO can trigger a graceful restart that preserves progress via the tracker.json checkpoint system. The restarted session picks up exactly where it left off.