sparkstation-diagnose

One-command health check for Sparkstation. Shows GPU status, model states, container health, recent log errors, memory usage, and thermal status. Use when asked to "diagnose", "check health", "what's wrong", "system status", or "why is X slow/broken".

跳到安装

仓库: kshetrajna12/sparkstation
最近来源活动: 2026年3月15日 21:26
检测到的 SKILL.md 语言: 英语
星标: 6
分支: 1

安装方式

默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

直接命令不会经过审查 Prompt；运行前请先检查来源。

npx skills add https://github.com/kshetrajna12/sparkstation --skill sparkstation-diagnose

命令会保持在同一行。复制前请横向滚动并检查完整内容。

想先保存到本地？可下载 SkillsMP 当前能够提供的文件。

检查来源文件

决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。

文件资源管理器

2 个文件

SKILL.md

来源说明 · 只读预览

返回搜索返回安装

sparkstation-deploy

Safe deployment operations for Sparkstation. Profile switching with drain/health-check, rolling model updates, Docker image rebuilds, full stop/start, and rollback. Use when asked to "deploy", "switch profile", "restart sparkstation", "update model", "rebuild container", or "rollback".

2026-03-16

sparkstation-bench

Benchmark inference throughput and latency for Sparkstation models. Measures TTFT, tokens/sec, p50/p95/p99 latency, and embedding throughput. Use when asked to "benchmark", "measure latency", "test throughput", or "profile inference performance".

2026-03-15

sparkstation-config-lint

Validate models.yaml profiles for correctness, consistency, and resource feasibility. Checks memory limits, alias collisions, required args, backend compatibility. Use when asked to "lint config", "validate models.yaml", "check config", or before deploying a new profile.

2026-03-15

查看这个仓库的全部 Skills

Sparkstation Diagnose

Comprehensive system health check in a single command.

Usage

# Full diagnostic report python3 .pi/skills/sparkstation-diagnose/scripts/diagnose.py # Quick status only (no log analysis) python3 .pi/skills/sparkstation-diagnose/scripts/diagnose.py --quick # JSON output for programmatic use python3 .pi/skills/sparkstation-diagnose/scripts/diagnose.py --json # Check specific component python3 .pi/skills/sparkstation-diagnose/scripts/diagnose.py --component gpu python3 .pi/skills/sparkstation-diagnose/scripts/diagnose.py --component models python3 .pi/skills/sparkstation-diagnose/scripts/diagnose.py --component containers python3 .pi/skills/sparkstation-diagnose/scripts/diagnose.py --component logs python3 .pi/skills/sparkstation-diagnose/scripts/diagnose.py --component gateway

What It Checks

GPU: Memory usage, temperature, power draw, utilization, thermal warnings

Supervisor: Health endpoint, running/failed/suspended model counts

Models: Per-model status, health, idle time, memory allocation vs actual

Containers: Docker container status, restarts, resource usage

Gateway: LiteLLM health, registered models, routing status

Logs: Recent errors/warnings from sparkstation.log (last 100 lines)

Resources: Memory allocation vs limits, port usage, disk space for models

Output

Color-coded terminal output with ✅/⚠️/❌ indicators:

═══ SPARKSTATION DIAGNOSTICS ═══ GPU STATUS ✅ OK Memory: 45.2 / 128.0 GiB (35.3%) Temperature: 52°C (threshold: 80°C) Power: 180W / 400W SUPERVISOR ✅ OK Status: healthy Models: 4 running, 0 suspended, 0 failed MODELS ✅ qwen3-vl-30b RUNNING port:8001 mem:55.0GB idle:42s ✅ bge-m3 RUNNING port:8002 mem:2.5GB idle:15s ✅ clip-vit RUNNING port:8003 mem:5.0GB idle:30s ❌ flux-dev FAILED port:8004 mem:35.0GB CONTAINERS ✅ sparkstation-qwen3-vl-30b Up 2 hours restarts:0 ✅ sparkstation-bge-m3 Up 2 hours restarts:0 ⚠️ sparkstation-flux-dev Exited (1) restarts:3 RECENT ERRORS (last 15min) [14:32:01] ERROR: Failed to health-check flux-dev: Connection refused [14:31:45] ERROR: Container sparkstation-flux-dev exited unexpectedly