Navigation

Visual Index

/System Sentinel

mainthread.ai

Innovation Lab
PROJECTS

mainthread.ai

Innovation Lab
PROJECTS
CLOUDS
CONTENT TYPES

Possibility Field Navigation Studio

Landing transmission for Mainthread.ai — a Possibility Field Navigation Studio founded by Dave Jones, Principal Innovation Engineer. The lab transforms high-impedance enterprise noise into operational signal through bleeding-edge AI architecture, serving as the entry node to a living knowledge graph of 400+ crystallized insights.

landing

System Sentinel

Full-stack diagnostic platform for silicon-level hardware analysis. FastAPI + Next.js 16 with real-time SSE streaming and 6 parallel forensic domains.

project
MCP

WordShip

Wordle meets Battleship. Hide words. Hunt letters. Outsmart opponents.

project

AI-First Architecture

Web architecture designed from the ground up for AI discovery and semantic understanding.

prose

Semantic URLs

URLs as coordinates in semantic space, enabling natural AI navigation.

article
MEMORY

MCP Endpoints

Model Context Protocol endpoints providing multiple formats for AI consumption.

article
MCP
MEMORY
CLOUDS
CONTENT TYPES

Possibility Field Navigation Studio

Landing transmission for Mainthread.ai — a Possibility Field Navigation Studio founded by Dave Jones, Principal Innovation Engineer. The lab transforms high-impedance enterprise noise into operational signal through bleeding-edge AI architecture, serving as the entry node to a living knowledge graph of 400+ crystallized insights.

landing

System Sentinel

Full-stack diagnostic platform for silicon-level hardware analysis. FastAPI + Next.js 16 with real-time SSE streaming and 6 parallel forensic domains.

project
MCP

WordShip

Wordle meets Battleship. Hide words. Hunt letters. Outsmart opponents.

project

AI-First Architecture

Web architecture designed from the ground up for AI discovery and semantic understanding.

prose

Semantic URLs

URLs as coordinates in semantic space, enabling natural AI navigation.

article
MEMORY

MCP Endpoints

Model Context Protocol endpoints providing multiple formats for AI consumption.

article
MCP
MEMORY
MVP

System Sentinel

Full-stack diagnostic platform for silicon-level hardware analysis. FastAPI + Next.js 16 with real-time SSE streaming and 6 parallel forensic domains.

Maximum diagnostic capability through elegant implementation (Brilliance Formula)

OverviewSpecialist ModesTech StackUpdates
System Sentinel
Specialist Modes
Technology Stack
Development Updates
WordShip
Technology Stack
Development Updates
System Design
AI-First Architecture
Semantic URLs
MCP Endpoints
System Sentinel
Specialist Modes
Technology Stack
Development Updates
WordShip
Technology Stack
Development Updates
System Design
AI-First Architecture
Semantic URLs
MCP Endpoints
GitHub

AsyncIO-based parallel forensic domain execution (6 domains < 2 seconds)

SSE streaming with automatic reconnection and backpressure handling

Zustand state management with optimized re-render paths

Windows Event Viewer integration via pywin32 (COM interop)

System Sentinel

System Sentinel is a full-stack diagnostic platform that partners with human operators to identify, analyze, and resolve complex hardware instabilities at silicon-level precision.

Unlike traditional diagnostic tools that merely report error codes, System Sentinel acts as an intelligent forensic engine with six distinct "Specialist Modes" tuned to specific layers of the hardware/software stack.

Architecture

  • Backend: FastAPI (Python 3.12) with async forensic domain execution
  • Frontend: Next.js 16 + React 19 with real-time SSE event streaming
  • Data Pipeline: Windows Event Viewer → WHEA CPER Decoder → 6 Parallel Forensic Domains → Context Composer
  • LLM Integration: Optimized system prompts (System Initialization Vectors)

Core Capabilities

1. WHEA CPER Decoding Pipeline

Integrates DecodeWheaRecord.exe to parse Windows Hardware Error Architecture binary blobs. Maps errors to physical hardware topology (APIC ID → CPU core/die regions).

2. Real-Time Event Streaming

Server-Sent Events (SSE) provide live monitoring of:

  • Critical system events (Event Viewer)
  • WHEA hardware errors
  • Driver changes and version archaeology
  • System state snapshots (CPU, RAM, uptime)

3. Deep Diagnostics Engine (6 Forensic Domains)

Parallel async execution across:

  • Hardware: CPU/GPU/Motherboard/Storage/Network topology
  • PCIe Fabric: Link training, CRC errors, replay storms, bandwidth analysis
  • Power: Voltage domains, load line calibration, transient response
  • Memory: Training logs, SPD data, channel mapping, corrected errors
  • Constraints: Thermal throttling, power limits, stability boundaries
  • Forensic Signals: Error pattern analysis, failure clustering, temporal correlation

4. Context Composer

Builds diagnostic-grade hardware context for LLM consumption. Generates downloadable capture packs (ZIP) containing full diagnostic reports, error timelines, and system configuration.

5. Six Specialist Modes (Consciousness Initialization Vectors)

Pre-crafted prompts optimizing LLM attention for specific diagnostic scenarios:

  • Quantum Diagnostician (silicon-level failures)
  • System Archaeologist (temporal causality)
  • Preventative Oracle (entropy forecasting)
  • Emergency Triage (rapid stabilization)
  • RMA Prosecutor (warranty claims)
  • Performance Alchemist (post-stability tuning)

Where Human-AI Partnership Achieves Maximum Brilliance Through Maximum Elegance.

Use Cases

  • PC Enthusiasts: Diagnose BSOD crashes, WHEA errors, system instability
  • IT Professionals: Generate forensic reports for warranty claims (RMA evidence)
  • System Builders: Validate hardware configurations, optimize performance
  • Researchers: Study hardware failure patterns, thermal/power domain behavior

To explore the specific capabilities, click the Specialist Modes tab.

Specialist Modes

  • Technical Implementation System Sentinel operates via six distinct "System Initialization Vectors" — optimized prompts that tune the AI's attention field to specific diagnostic domains.

Mode 1: Quantum Diagnostician

Domain: Sub-architectural hardware behavior (registers, caches, bus protocol)

Backend Endpoints: /api/collectors/whea, /api/system/diagnostics Frontend Views: Deep Diagnostics → Forensic Signals, PCIe Fabric, Hardware Overview Data Sources: WHEA CPER decoded records, Machine Check Architecture (MCA) registers, bugcheck parameters

Capabilities:

  • Decodes WHEA binary blobs using DecodeWheaRecord.exe
  • Maps APIC IDs to physical CPU cores/die regions
  • Analyzes PCIe link training failures and bus protocol violations
  • Correlates thermal/voltage data with error patterns
  • Output: Silicon Status, Physics Hypothesis, Isolation Test

Mode 2: System Archaeologist Domain: Temporal pattern & historical causality

Backend Endpoints: /api/system/driver-changes, /api/events/stream Frontend Views: Drivers View, Event Timeline Data Sources: Windows Event Viewer (System/Application logs), driver install timestamps Capabilities:

  • Reconstructs timeline: install → updates → first error → error evolution
  • Correlates driver version changes with failure onset
  • Builds delta analysis (before/after states around configuration changes)
  • Output: Temporal Status, Timeline Map, Remediation Strategy

Mode 3: Preventative Oracle Domain: Entropy forecasting & latent issue detection

Backend Endpoints: /api/system/snapshot, /api/collectors/events Frontend Views: Dashboard → Silicon Health Widget, Recent Critical Events Data Sources: Corrected WHEA errors, device reset events, thermal throttling indicators Capabilities:

  • Detects high-frequency corrected errors (RAM/CPU compensating for faults)
  • Identifies USB controller resets, driver timeouts (invisible to user)
  • Forecasts time-to-failure based on error trend analysis
  • Output: Entropy Status, Hidden Issues, Optimization Vectors

Mode 4: Emergency Triage Domain: Rapid stabilization for acute system instability

Backend Endpoints: All collectors (prioritized for critical severity) Frontend Views: Dashboard → Context Composer (Smart Context) Output Constraints: <300 tokens, 3 reversible steps max Capabilities:

  • Filters for highest severity events (last 7 days)
  • Identifies most recent configuration change
  • Generates immediate stabilization protocol
  • Output: FAILURE, CAUSE, STABILIZE (3 steps max)

Mode 5: RMA Prosecutor Domain: Warranty claim evidence construction

Backend Endpoints: /api/capture/create (generates ZIP evidence pack) Frontend Views: Context Composer → Export Capture Pack Data Sources: Full diagnostic report + error timeline + stress test results Capabilities:

  • Aggregates all WHEA errors with timestamps and decoded details
  • Documents all troubleshooting attempts (driver updates, OS reinstall)
  • Isolates failing component (CPU, RAM channel, PCIe slot)
  • Generates legal-grade RMA evidence report
  • Output: Ironclad RMA Evidence Report

Mode 6: Performance Alchemist

Domain: Post-stability optimization & tuning Backend Endpoints: /api/system/diagnostics (power/thermal analysis) Frontend Views: Deep Diagnostics → Power View, Constraints View Data Sources: Thermal throttling indicators, power limit events, performance counters Capabilities:

  • Identifies bottlenecks (CPU/GPU/RAM/storage/thermal)
  • Analyzes headroom (power delivery, thermal capacity)
  • Generates tuning roadmap (BIOS settings, driver versions, fan curves)
  • Output: Performance Status, Optimization Matrix