# rai

> rai — classevelabs-rai. Use this tool when you need to efficiently run large language models (LLMs) on CPU-only systems, solving problems like fast inference and local deployment without relying on GPUs or Python runtime. It takes in quantized models and input data, outputting generated text through a local HTTP/MCP server interface. Ideal for use cases requiring low-latency, CPU-based LLM inference in resource-constrained environments.

Canonical page: https://skillsregistry.net/skills/classevelabs-rai  
JSON: https://api.skillsregistry.net/v1/skills/classevelabs-rai

## Description

CPU-only LLM inference engine in pure Rust — 4-bit quantized models, hand-written AVX2 kernels, speculative decoding, and a local HTTP/MCP server. No GPU, no Python runtime.

## Trust

- **Trust score (0–1):** 0.50
- **Verification tier:** unverified

## Facts

- **Version:** 1.0.0
- **Skill type:** atomic
- **Execution layer:** container
- **Runtime environment:** vm
- **Updated:** 2026-09-19

## Source

- **Source listing:** [GitHub](https://github.com/Classevelabs/rai)

## Use it

Resolve this record through the SkillsRegistry MCP server (no auth, read-only):

```
claude mcp add --transport http --scope user skillsregistry https://api.skillsregistry.net/mcp
```

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_skill",
    "arguments": {
      "slug": "classevelabs-rai"
    }
  }
}
```

REST: `GET https://api.skillsregistry.net/v1/skills/classevelabs-rai` · pull for local use: `GET https://api.skillsregistry.net/v1/skills/classevelabs-rai/pull`

---
SkillsRegistry indexes agent skills from public registries and GitHub. Skills we have analysed are scanned with Circle-IR and scored on six dimensions; each listing states its scan coverage. More: https://skillsregistry.net/llms.txt