spectredeath/skill-flywheel/skill-model-health-monitor
Claude Code and agent skill by spectredeath/skill-flywheel
- community
Checking that the SKILL.md file is still in the repository.
Install
npx vskill@latest install spectredeath/skill-flywheel/skill-model-health-monitorSKILL.md
The same SKILL.md was published earlier as spectredeath/skill-flywheel/skill-model-health-monitor.
origin: manual triggers:
- agent
- ai
- development
quality: applied_count: 0 success_count: 0 completion_rate: 0.0 token_savings_avg: 0.0 created_at: "2026-03-24T10:00:00Z" updated_at: "2026-03-24T10:00:00Z"
Model Health Monitor
Description
A diagnostics engine that performs automated health checks on model endpoints by evaluating GPU memory, inference errors, and token throughput stability.
Purpose
Ensures only healthy models are included in active missions, automatically flagging and decommissioning degraded endpoints.
Capabilities
- VRAM utilization tracking
- Inference error-code auditing (429, 500, 503)
- Throughput performance degradation detection
- Self-healing trigger generation (Context reset, Service restart)
Usage Examples
- "Check health of Llama-70B on Port 8012" -> Status: Healthy (VRAM 18GB/24GB)
- "Audit error rates for GPT-4o" -> Status: Degraded (Rate 5.4%)
Input Format
Model ID or endpoint URL to audit.
Output Format
Health status (Healthy, Degraded, Critical), VRAM stats, and error log summary.
Configuration Options
error_threshold_percentage: Default 1.0%vram_warning_level: Default 85%
Constraints
- ALWAYS perform a health check before including a model in a high-priority mission.
- MUST flag models as "Critical" if error rate exceeds 5.0%.
- MUST release VRAM in 30s if model is idle and flagged.
Examples
Example 1: GPU Under Stress
Input: Llama-3 (Port 8000) Output: Status: Degraded (VRAM 23.5GB/24GB)
Error Handling
- Case: NVML Failure -> Action: Use basic inference timing for health estimation.
- Case: No Probe Output -> Action: Flag as "Unknown/Stalled".
Performance Optimization
- Use passive monitoring of active requests to reduce probe overhead.
- Aggregate metrics in a shared
performance_monitorsingleton.
Integration Examples
- Integrated with
discovery_service.pyto update service status. - Integrated with
McpHealthCheckin Docker.
Best Practices
- Monitor token throughput (Tokens/Sec) for "Silent Degradation".
- Use binary state checks (UP/DOWN) for fast routing decisions.
Troubleshooting
- Symptom: Constant False Positives -> Check: Thresholds too sensitive.
- Symptom: Stale Health Data -> Check: Metric TTL in memory.
Monitoring and Metrics
- Track
model_uptime_percentage. - Monitor
average_recovery_time_ms.
Dependencies
performance_monitornvidia-ml-py
Version History
- 1.0.0: Initial release
- 4.2.0: Synchronized with Skill Flywheel industrial standard
License
MIT License
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingdomains/model_orchestration/SKILL.model-health-monitor/SKILL.md
Popularity
- Repository stars
- 1
- Repository forks
- 1
- 7-day trend score
- 8
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 28 Oct 2026, 21 days ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/spectredeath/skill-flywheel/skill-model-health-monitor)