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/yamaru-eu/hardware-probe/local-inference-optimizernpx skills add yamaru-eu/hardware-probe --skill local-inference-optimizergit clone --depth 1 https://github.com/yamaru-eu/hardware-probeWrote 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/yamaru-eu/hardware-probe/local-inference-optimizer)<a href="https://agentmods.dev/skills/yamaru-eu/hardware-probe/local-inference-optimizer"><img src="https://agentmods.dev/badge/skills/yamaru-eu/hardware-probe/local-inference-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 5.1 | $0.00023 | $0.00714 |
| Opus 5 | $0.00012 | $0.00357 |
| Sonnet 5 | $0.00005 | $0.00143 |
| Haiku 4.5 | $0.00002 | $0.00071 |
Grade A, and why
local-inference-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 6d 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 — 53 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill: Local Inference Optimizer
Expert system for analyzing local hardware/software topology and providing actionable recommendations for LLM inference (Ollama, LM Studio, vLLM).
Context
Use this skill when a user wants to:
- Optimize their machine for local LLMs.
- Troubleshoot slow inference speeds.
- Understand why a model isn't using the GPU.
- Calculate if a specific model/quantization will fit.
Diagnostic Protocol
1. Memory Bandwidth Analysis
- Rule: Inference speed is directly bound by Memory Bandwidth (GB/s).
- Heuristics:
- < 20 GB/s: Slow (CPU/Single Channel DDR4). Recommend small models (3B) or hardware upgrade.
- 20-60 GB/s: Standard (Dual Channel DDR4/DDR5). Good for 7B-14B models.
-
100 GB/s: High Performance (Apple M-Series, Quad Channel, or High-end GPU VRAM).
- Advice: If
memory_bandwidth_gbsis significantly lower than theoretical specs, suggest checking RAM slots (Dual Channel) or BIOS XMP/EXPO profiles.
2. VRAM & Quantization Strategy
- Rule: Model + KV Cache must fit in VRAM for 10x speedup.
- Formulas:
ModelSize = (Params * Quant) / 8(e.g., 7B @ 4-bit ≈ 3.5GB).KVCache = ContextLength * Params * 0.0000006(Rough estimate).
- Optimization: If VRAM is tight, suggest reducing
num_ctx(context length) before dropping quantization quality.
3. Runtime Troubleshooting (WSL/Docker/Ollama)
- WSL2: If
memory_limit_gbin.wslconfigis < 50% of total RAM, suggest increasing it. - Docker: If
hasNvidiaRuntimeis false, provide thenvidia-container-toolkitinstallation steps. - Environment: If
OLLAMA_NUM_PARALLELis missing, suggest setting it to 2 for multi-agent workflows.
4. Hardware Health (Thermals & Storage)
- Thermals: If inference starts fast but slows down over time, call
thermal_profileto check for frequency clipping or overheating. - Storage: If model loading is slow, call
check_storage_healthto verify if the model is on a slow HDD vs NVMe SSD.
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.
- 6d ago First seen · 53 lines · 23 tokens per session scan A ab731d5d14aa
local-inference-optimizer is a skill published in the GitHub repository yamaru-eu/hardware-probe (7 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 23 tokens to every session and 714 once invoked, about $0.0001 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.