Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Local Llm Efficiency

ASecurity

Run huge models locally via quantization (GGUF/GPTQ/AWQ). Use when 740B in 25GB (Colibri) or need 1000x cheaper local inference with llama.cpp/Ollama. Triggers on \"Colibri\", \"quantizacao\", \"GGUF\", \"local llm\", \"ollama\", \"vLLM local\"

2 stars
0 votes
0 copies
0 views
Added 9/19/2026
ai-agentsgo

Security Analysis

A100/100

Scanned 9/19/2026

Install to Claude Code

$npx -y skills add majinmagros/magros.ai-skills --skill local-llm-efficiency --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Local Llm Efficiency?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Local Llm Efficiency
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/majinmagros-local-llm-efficiency/badge)](https://www.skillsdirectory.com/skills/majinmagros-local-llm-efficiency)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: local-llm-efficiency
description: "Run huge models locally via quantization (GGUF/GPTQ/AWQ). Use when 740B in 25GB (Colibri) or need 1000x cheaper local inference with llama.cpp/Ollama. Triggers on \"Colibri\", \"quantizacao\", \"GGUF\", \"local llm\", \"ollama\", \"vLLM local\""
---

# Local LLM Efficiency — Colibri 740B@25GB

> Fonte: `Maestros da IA — 1p0HXLv_5wM` (PANIC, Colibri), transcript `1p0HXLv_5wM.pt.dedup.txt:40-120`

Cloud pay-to-play vs local 1000x cheaper, lento/experimental mas movimento open-source.

## Quando usar

- Rodar modelo gigante em consumer hardware (25-32GB RAM)
- Custo cloud inviável, precisa offline/grátis
- Validar claim 740B@25GB antes de comprar hardware

## Técnicas

| Método | Ferramenta | Trade-off |
|---|---|---|
| GGUF | llama.cpp, Ollama, LM Studio | CPU, lento, 100% offline |
| GPTQ/AWQ/EXL2 | AutoGPTQ, vLLM | GPU, mais rápido |
| Speculative decoding | vLLM | + velocidade |

## Workflow

1. Escolha modelo + quant (ex: `Q4_K_M` → 25GB)
2. `ollama run <model>:q4` ou `llama.cpp --model model.gguf`
3. Benchmark: tokens/s, RAM, qualidade vs cloud (use `agent-eval`)
4. Decida: cloud (veloz) vs local (barato) — veja `local-ai-hardware` para TCO

## Checklist

- [ ] Modelo + quant validados
- [ ] Benchmark local vs cloud
- [ ] Fonte primária Colibri verificada (anti-hallucination)

## Tese independencia (Batch 17a, #50 #53)

Nuvem = aluguel do modelo + entrega do seu IP (pay-twice); governo ou
vendor pode cortar seu acesso (caso Fable/export-ban). Local open source
= privacidade total, custo de eletricidade, controle e offline. Direcao:
velocidade, custo e memoria — quando o local ficar bom o bastante, a
vantagem da nuvem vira so conveniencia.

## Referências

## Exemplo

```text
Claim "740B em 25GB" → valida: Q4_K_M via Ollama/llama.cpp no hardware real
Benchmark: tokens/s + RAM + qualidade vs cloud (agent-eval); fonte Colibri checada
Resultado: lento mas 1000x mais barato → vale p/ lote offline, não p/ interativo
```

Attribution

majinmagrosmajinmagros
View sourceMore from majinmagros →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.

1023331 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3331 votes

catchup

Recovers prior coding-agent session context by running `catchup <agent> --since-compact`, which extracts a clean summary of a previous Codex, Claude Code, Antigravity, OpenCode, or Pi Agent session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", or asks to recover/summarize a previous session before continuing. Do NOT use for the current conversation, git history, or any non-agent log.

611 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →