Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/chf3198/copilot-governance/fleet-model-optimizernpx skills add chf3198/copilot-governance --skill fleet-model-optimizergit clone --depth 1 https://github.com/chf3198/copilot-governanceWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/chf3198/copilot-governance/fleet-model-optimizer)<a href="https://agentmods.dev/skills/chf3198/copilot-governance/fleet-model-optimizer"><img src="https://agentmods.dev/badge/skills/chf3198/copilot-governance/fleet-model-optimizer.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00041 | $0.00927 |
| Opus 5 | $0.00020 | $0.00464 |
| Sonnet 5 | $0.00008 | $0.00185 |
| Haiku 4.5 | $0.00004 | $0.00093 |
Grade A, and why
fleet-model-optimizer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 85 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Fleet Model Optimizer Skill
Use this skill to audit and improve Ollama model selections across a device fleet. Invoke when models feel stale, performance is low, or after a major LLM release cycle.
Scope Boundary
Owns: hardware-to-tier mapping, model scoring, pull/delete command generation.
Hands off to: Admin for actual ollama pull/delete execution on remote devices.
Hard Constraints
- Never reference specific device IPs, hostnames, or user credentials.
- Never recommend proprietary closed-weight models — Ollama registry only.
- Always produce a pull-before-delete transition plan (no service gaps).
- Respect RAM budget: leave ≥20% free after all models loaded.
- Verify Ollama registry availability before recommending any model.
Tier Classification
Classify each device from merged fleet inventory (node scripts/global/fleet-config.js fleet) by available RAM:
| Available RAM | Tier | Max model size | Examples |
|---|---|---|---|
| < 1.5 GB | slm | ~500 MB | gemma3:270m, qwen3.5:0.5b |
| 1.5–4 GB | slm+ | ~1.5 GB | gemma4:e4b, qwen3.5:1.5b |
| 4–10 GB | mid | 7–8 B | qwen3:8b, mistral-nemo:12b* |
| 10–20 GB | mid+ | 8–14 B | qwen3:8b, phi4:14b |
| 20+ GB | high | 30–70 B | deepseek-r1:32b, qwen3:30b |
*Only if swap or GPU offload is available.
Step-by-Step Execution
1. Inventory Read
Run node scripts/global/fleet-config.js fleet (or read ~/.megingjord/devices.json overlay). For each device extract:
id, ram.available, gpu (if present), ollamaModels, ollamaWarmTokPerSec.
2. Tier Classification
Apply the table above to assign each Ollama-enabled device a tier. Note any GPU presence — it enables larger models and faster inference.
3. Model Scoring (per tier)
Search Ollama registry trends (pull counts, recency) for top candidates:
- Prioritize: reasoning quality, coding capability, pull count > 1M (community validation).
- Include 1 primary + 1–2 fallback models per device.
- Flag models with < 100K pulls as experimental.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 85 lines · 41 tokens per session scan A a35dcf62a2e0
fleet-model-optimizer is a skill published in the GitHub repository chf3198/copilot-governance (1 stars, last pushed 1mo ago), licensed MIT. It adds 41 tokens to every session and 927 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
integrated-browser
Use this when working on the VS Code integrated browser ("browserView") to understand its architecture and mental model. Covers the embedded Chromium browser, its editor tab, navigation, overlay/layout, sessions, and agent browser tools under src/vs/platform/browserView and src/vs/workbench/contrib/browserView.
spark-environment-setup
Set up a working ML training/inference environment on NVIDIA DGX Spark (GB10, aarch64, CUDA 13). Use when installing PyTorch/Unsloth/TRL/vLLM on DGX Spark, hitting libcudart or wheel-ABI errors on aarch64, or choosing between NGC containers and bare pip installs.
spark-memory-thermal-ops
Manage unified memory and thermals during long-running ML jobs on NVIDIA DGX Spark. Use when planning memory headroom for a training run on GB10, when a job OOMs on unified memory, or when monitoring temperature and power during multi-hour training.
spark-training-gotchas
Preflight and diagnose the ten known failure modes for ML training on NVIDIA DGX Spark. Use when a training run on DGX Spark fails to start, OOMs below the 128GB limit, slows down mid-run, or before any multi-hour training job on GB10.
llama-cpp
Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.
amc-run-rtsp-calibration
Calibrate a new dataset from live RTSP camera streams via the AutoMagicCalib REST API. Use when the user provides RTSP URLs or asks to calibrate live cameras; VIOS records clips, AMC ingests them, then runs calibration.