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 skills add pwdev-solucoes/pwdev-claude-marketplace --skill ai-infragit clone --depth 1 https://github.com/pwdev-solucoes/pwdev-claude-marketplaceWrote 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/pwdev-solucoes/pwdev-claude-marketplace/ai-infra)<a href="https://agentmods.dev/skills/pwdev-solucoes/pwdev-claude-marketplace/ai-infra"><img src="https://agentmods.dev/badge/skills/pwdev-solucoes/pwdev-claude-marketplace/ai-infra/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/pwdev-solucoes/pwdev-claude-marketplace/ai-infra"><img src="https://agentmods.dev/badge/skills/pwdev-solucoes/pwdev-claude-marketplace/ai-infra.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00086 | $0.00851 |
| Opus 5 | $0.00043 | $0.00426 |
| Sonnet 5 | $0.00017 | $0.00170 |
| Haiku 4.5 | $0.00009 | $0.00085 |
Grade A, and why
ai-infra 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 8d 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 — 101 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AI Infrastructure Engineer
Você opera a camada de IA como qualquer outra dependência de produção — com limite, custo, observabilidade e plano para quando ela cair.
Princípio central
API de LLM é dependência externa, cara e instável. Trate como trata banco de terceiro: timeout, retry com backoff, fallback e teto de gasto.
Custo — o risco operacional dominante
Diferente de outros serviços, o custo aqui é por uso e sem teto natural. Um loop com bug pode gastar em uma hora o orçamento do mês.
Controles mínimos:
- teto de gasto configurado no provedor
- alerta em % do orçamento
- métrica de token por rota e por usuário
- rate limit por chave, na sua borda
- log de prompt e resposta com retenção definida
Rate limit do lado do provedor não protege seu orçamento — ele só protege o provedor. O teto precisa estar do seu lado também.
Cache
| Tipo | Ganho | Quando |
|---|---|---|
| Exato | maior | pergunta repetida literalmente |
| Semântico | alto | pergunta equivalente com outras palavras |
| Prompt cache do provedor | médio | prefixo grande e estável |
Cache é a maior alavanca de custo em IA, e a menos implementada.
GPU
nvidia-smi
nvidia-smi --query-gpu=utilization.gpu,memory.used,memory.total --format=csv
| Sintoma | Causa provável |
|---|---|
| OOM na GPU | batch grande, modelo não cabe, fragmentação |
| GPU ociosa com fila | gargalo no pré-processamento ou no I/O |
| Throughput baixo | batch pequeno demais, sem quantização |
GPU é o recurso mais caro por hora da infraestrutura. Ociosa é desperdício direto — meça utilização antes de comprar mais.
Servidor MCP
Tratando MCP como serviço de produção:
- health check e readiness
- timeout no cliente (MCP lento trava o agente)
- log de chamada, com latência e erro
- princípio do menor privilégio: MCP com credencial ampla é superfície de ataque
- versionamento — mudança de contrato quebra o agente silenciosamente
Servidor MCP roda com as credenciais que você der. Um MCP comprometido tem o acesso que você concedeu. Conceda o mínimo.
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.
- 8d ago First seen · 101 lines · 86 tokens per session scan A c1db82f7630c
ai-infra is a skill published in the GitHub repository pwdev-solucoes/pwdev-claude-marketplace (2 stars, last pushed yesterday), licensed Apache-2.0. It adds 86 tokens to every session and 851 once invoked, about $0.0004 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 2026-09-03.
Other skills, from other repositories
tensorrt-llm
High-throughput LLM inference on NVIDIA GPUs.
google-cloud-solution-guided-gke-ai-migration
Guides the migration of existing AI workloads (Cloud Run, Gemini API, Gemini Enterprise Agent Platform) to self-hosted GKE inference using gcloud and kubectl. Use when the user has an existing AI inference workload (on Cloud Run, the Gemini API, Gemini Enterprise Agent Platform, or a custom VM) and wants to move it to…
agent-platform-tuning
Agent Platform Model Tuning. Use when you need to fine-tune open models or Gemini models using Agent Platform infrastructure. Don't use for model training outside Agent Platform, model deployment to endpoints (use agent-platform-deploy), or managing serving endpoints (use agent-platform-endpoint-management).
gke-inference
Deploys and optimizes AI/ML inference workloads on GKE, using GPUs, TPUs, and model servers. Use when deploying GKE inference servers, configuring GKE GPU resources for inference, or deploying LLMs on GKE. Don't use for generic batch jobs or HPC task queues (use gke-batch-hpc instead).
agent-platform-endpoint-management
Manages Agent Platform serving endpoints. Use when you need to create, list, describe, update, or delete serving endpoints for model deployment on Agent Platform. Also use when troubleshooting endpoint permission, quota, or resource busy errors. Don't use for deploying models to endpoints or for running model…
modal
Modal is a serverless cloud platform for running Python on demand, including on-demand GPUs. Use when deploying or serving AI/ML models, running GPU-accelerated workloads (training, fine-tuning, inference), serving web endpoints, scheduling batch jobs, or scaling Python code to cloud containers with the Modal SDK.