CORTEX OK JUNIPER 9 CRONS MORTEM 60s RELAY ALERT THRESHOLD CPU 23% MEMORY 61% NETWORK 2.3 MB/s UPTIME 99.97% [03:42:11] SENTINEL OK — all agents nominal [03:41:09] MORTEM relay confirmed — tx: 4Kxq7R... [03:40:55] THRESHOLD ALERT — metric spike, auto-acknowledged ALWAYS-ON — 2 DEPLOYMENTS — THRESHOLD ESCALATION
Always-On Monitoring Agent
SENTINEL
Two simultaneous deployments. CORTEX and Juniper. Threshold-based escalation. If it matters, SENTINEL surfaces it.
Architecture Docs Research Notes
Live — 2 Deployments
2
Live deployments
Always
On 24/7
Threshold
Escalation
Scroll
Celaya Solutions / Research Ecosystem
CORTEXCLOSEPPEMORTEMProject JupiterNeural ChildCelaya Chain ProtocolJuniperMERIDIANSIGNALVERDICTSENTINELCORTEXCLOSEPPEMORTEMProject JupiterNeural ChildCelaya Chain ProtocolJuniperMERIDIANSIGNALVERDICTSENTINEL
SENTINEL.
In detail.
2
Simultaneous Deployments
SENTINEL runs in CORTEX and Juniper simultaneously. Two separate monitoring contexts, two separate threshold configurations, one consistent behavioral contract.
Always-On
Continuous Monitoring
SENTINEL does not run in sessions. It monitors continuously. The threshold configuration defines what triggers escalation. Everything below threshold runs silently.
Threshold
Escalation Logic
When a metric crosses a threshold, SENTINEL escalates. The escalation target depends on context: CORTEX alerts go to floor stakeholders, Juniper alerts surface to the researcher.
Silent
Default Mode
The design goal: SENTINEL should be invisible until something requires attention. Noise is a failure mode. Signal is the deliverable.
Shared
Architecture
The same SENTINEL architecture runs in both deployments. Different configurations, identical contract. A monitoring agent that works the same way in any context.
Embedded
In CLOS
SENTINEL is also one of the 37 CLOS agents, where it monitors the cognitive operating system rather than manufacturing or personal automation systems.
If everything is an alert,
nothing is.

SENTINEL is designed around the principle that monitoring agents should be invisible until they are not. The threshold configuration is the design. Get it wrong and you have noise. Get it right and you have a system that tells you exactly what you need to know, exactly when you need to know it.

Engage the research.

This is an active research instrument, not a product brochure. Collaboration on the framework, methodology, and architecture is open.

01
Research Collaboration
Architecture, methodology, open questions
02
Architecture Notes
Technical documentation and system design
03
Contact Research Lab
System Status
Live — 2 Deployments
Status
Live — 2 Deployments
Lab
Celaya Solutions
Location
El Paso, TX
The record.
CORTEX
First deployment.
SENTINEL was first built as part of the CORTEX manufacturing intelligence system. The monitoring requirements for a 14-agent manufacturing platform defined the initial architecture.
Juniper
Second deployment.
When Juniper was built with its 9 cron jobs and 7 agents, SENTINEL was adapted for the personal automation context. Same architecture, different configuration.
CLOS
Third context.
SENTINEL integrated into the CLOS 37-agent architecture as the dedicated monitoring agent. Now watching a cognitive system rather than a manufacturing or automation system.
Now
Three active contexts.
CORTEX, Juniper, and CLOS. Three deployments, one consistent architecture. SENTINEL is the most widely deployed instrument in the Celaya Solutions ecosystem.