A hardware-aware advisor that detects system specs to recommend the best-fitting local LLMs with optimal quantization and speed estimates.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install llmfit
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install llmfit using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is llmfit-advisor?
The llmfit-advisor is a technical utility designed to bridge the gap between local hardware capabilities and Large Language Model requirements. By analyzing system resources like RAM, CPU, and VRAM across NVIDIA, AMD, and Apple Silicon architectures, this tool ensures users select models that perform optimally without crashing. As part of the Openclaw Skills ecosystem, it provides data-driven insights into which models offer the best balance of quality and speed for your specific machine.
This skill eliminates the guesswork involved in setting up local inference. It calculates a composite score for models based on quality, speed, and memory fit, allowing developers to make informed decisions before downloading dozens of gigabytes of model weights. Whether you are using Ollama, vLLM, or LM Studio, this skill streamlines the configuration process.
llmfit-advisor Use Cases
- Identifying which local models are compatible with specific GPU VRAM limits
- Recommending optimal quantization levels for models like Llama 3.1 or Mistral
- Configuring local inference providers like Ollama or LM Studio within Openclaw Skills
- Finding the best coding or reasoning models based on available system memory
- Estimating tokens per second (TPS) before downloading large model files
How llmfit-advisor Works
- The skill executes a system scan to detect CPU cores, total RAM, and GPU details including VRAM and backend support.
- It queries the recommendation engine to compare hardware specs against a library of model requirements.
- A ranking is generated using a composite score that weighs model quality against performance metrics.
- The skill identifies the 'Best Quant' (optimal quantization) that fits within the detected memory profile.
- Resulting models are mapped to specific provider tags, such as Ollama identifiers, for easy installation.
llmfit-advisor Setup
To use this skill, you must have the llmfit binary installed. You can install it via Homebrew or Cargo:
# Install via Homebrew
brew tap AlexsJones/llmfit
brew install llmfit
# Or install via Cargo
cargo install llmfit
Once installed, Openclaw Skills can trigger the advisor whenever hardware-specific questions are asked.
llmfit-advisor Data Schema & Taxonomy
The skill organizes system and recommendation data into structured formats for clear decision making:
| Component | Details Captured |
|---|---|
| Hardware Specs | CPU name, total RAM, GPU VRAM, and unified memory status for Apple Silicon |
| Model Metrics | Parameter count, fit level (Perfect/Good/Marginal), and run mode (GPU/CPU) |
| Performance | Estimated tokens per second (TPS) and memory utilization percentage |
| Mappings | HuggingFace IDs mapped to local provider tags for Openclaw Skills configuration |
llmfit-advisor Advanced Features
- Multi-GPU support with automatic VRAM aggregation across multiple graphics cards
- Automated configuration of openclaw.json to set primary and fallback local models
- Use-case filtering to specifically find models optimized for coding, reasoning, or chat
- Deep hardware detection for NVIDIA (nvidia-smi), AMD (rocm-smi), and Apple Metal backends
- Minimum fit level filtering to exclude models that would result in poor performance
SKILL.md
Loading
Related Openclaw Skills
LLM Router
An intelligent proxy that classifies incoming requests by complexity and routes them to the most cost-effective LLM provider.
Caesar Research
Conduct multi-source deep research with verifiable citations and interactive follow-up chat via the Caesar API.
Timemap - Tel Aviv & Haifa Venue History
A historical mapping tool to explore the legacy of entertainment, nightlife, and cultural venues in Tel Aviv and Haifa.
StrikeRadar
A real-time OSINT monitoring tool that aggregates data across eight signal categories to estimate US-Iran military strike probability.
Prompt Assemble
A standardized prompt assembly framework designed to guarantee AI agent stability by preventing token overflow through intelligent memory orchestration.
Daily Insight Brief
An automated morning intelligence agent that collects, verifies, and distills high-impact business, technology, and financial news from premium global sources.
Featured*
