Reading
A marketplace featuring over 40 installable reference skills covering areas such as vLLM/SGLang inference, Kubernetes & Harvester, GPU host deployment, observability, security, and Agent workflows.
Section 01
A marketplace featuring over 40 installable reference skills covering areas such as vLLM/SGLang inference, Kubernetes & Harvester, GPU host deployment, observability, security, and Agent workflows.
Section 02
Section 03
air-gapped/skills is a plugin marketplace specifically designed for Claude Code, featuring over 40 carefully polished reference skills. These skills cover all aspects of enterprise-grade tech stacks, from AI model inference and Kubernetes cluster management to GPU host deployment and security hardening. For engineers using Claude Code for development and operations work, this is an extremely valuable resource library.
The project's uniqueness lies in the depth and professionalism of its skills. Unlike general code snippets or tutorials, each skill is a production-proven complete solution, including detailed operation guides, best practices, and troubleshooting methods. Whether you need to quickly deploy a vLLM inference service or plan a Harvester HCI upgrade path, you can find corresponding expert-level guidance.
Section 04
In the field of generative AI infrastructure, this project provides one of the most comprehensive skill sets in the industry:
vLLM Skill Suite: Contains 14 vLLM-related reference skills, covering everything from basic deployment and performance tuning to advanced features like prefix caching, speculative decoding, and multimodal inference.
SGLang Skill Suite: Also provides full SGLang support, including advanced features such as HiCache layered KV caching, Model Gateway routing, and multi-node distributed inference. As an emerging inference framework, SGLang excels in throughput and latency optimization.
NVIDIA Ecosystem Support: For NVIDIA data center GPU hosts, it provides a complete skill chain from hardware initialization and driver installation to NCCL network optimization. This includes the bring-up process for HGX/DGX series servers, Secure Boot configuration, and NVLink/NVSwitch network tuning.
Section 05
Harvester HCI Management: Provides detailed upgrade guides for Harvester hyper-converged infrastructure, covering the complete migration path from EOL versions to the latest stable versions, including coordinated upgrade strategies for components like RKE2, KubeVirt, and Longhorn.
Rancher Multi-Cluster Management: For upgrade planning of the community edition Rancher, especially for air-gapped multi-cluster fleet management scenarios, it provides verified upgrade sequences and risk control methods.
KEDA Autoscaling: In-depth explanation of Kubernetes Event-driven Autoscaling configuration and optimization, including the use of over 70 trigger types, multi-trigger combination strategies, and scale-to-zero production practices.
Argo CD GitOps: Covers complete skills for writing Applications and ApplicationSets, multi-source configuration, synchronization strategy optimization, and deployment methods like Helm/Kustomize/OCI.
Section 06
Prometheus & Grafana Mimir: Not only includes basic PromQL query writing, but also covers Mimir's distributed storage architecture, alert rule optimization, and management strategies for large-scale metric data.
GPU Performance Analysis: The NVIDIA AIPerf skill provides vendor-neutral benchmarking methods for generative AI inference, supporting multiple test modes such as concurrency testing, request rate testing, and fixed scheduling trace replay, covering 15 endpoint types.
Section 07
Secure Boot Certificate Rotation: Addresses the UEFI certificate rotation issue for Microsoft Secure Boot from 2011 to 2023, providing complete remediation solutions for Dell PowerEdge/iDRAC9 bare-metal servers, Ubuntu/Linux systems, and Harvester HCI/KubeVirt virtual machines.
Keycloak Identity Management: Covers Keycloak deployment, configuration, security hardening, and integration, supporting modern Quarkus distributions and Operator mode, suitable for building enterprise-grade IAM solutions.
Threat Modeling: Provides three modes of threat modeling skills—interview mode to guide application owners through the four-question framework, guided mode to derive threat models from code and historical data, and independent mode to provide a complete modeling workflow for security teams.
Section 08
Jinja Template Expert: Covers three major Jinja application scenarios: HuggingFace chat_template, Ansible playbook, and Kubernetes templates, helping developers write and debug complex template logic.
Helm Chart Development: From writing Chart.yaml and validating values.schema.json to generating helm-docs documentation, it provides complete best practices for Helm chart development, covering Helm 4's new features like SSA, WASM, and OCI digest support.
Makefile Best Practices: For GNU Make 4.x scenarios such as task orchestration, parallel safety, and self-documenting help targets, it provides reusable templates and patterns.
Keep going with more reads from the same topic.
Nornir MCP Server is an enterprise-level server based on the Model Context Protocol (MCP). It seamlessly integrates large language models (such as Claude) with the Nornir network automation framework, supporting natural language orchestration for multi-vendor network devices (Cisco, Arista, Juniper, etc.), and providing production-grade features like a dual-engine architecture (NAPALM + Netmiko), intelligent filtering, and a secure sandbox.
Bibliothèque Française LLM is a structured indexing and annotation project for French public domain literature designed specifically for large language models (LLMs). It integrates multiple authoritative sources such as DraCor, Common Corpus, and Wikisource, providing metadata indexing categorized by genre, author, and era, as well as in-depth annotations for dramatic texts (including characters, lines, stage directions, etc.). Its aim is to enable LLMs to efficiently read and understand classic French literary works.
Splinter is a minimalist, high-performance key-value (KV) and vector storage system enabling zero-latency inter-process communication via shared memory and atomic operations. With only 766 lines of core code, it supports millions of operations per second and 768-dimensional vector storage, offering a new architectural approach for local LLM inference and data-intensive applications.
A complete cloud-native AI application development guide for beginners, building a simple generative AI chatbot using Amazon EC2, Apache, Python CGI, and Amazon Bedrock, covering architecture design, IAM permission configuration, security best practices, and cost optimization suggestions.