Awesome-omni-skill cloud-native-engineer

The definitive skill for building and deploying high-performance, distributed systems using Cloud Native standards (Dapr, Redis, Microservices). Use when a project requires professional-grade architecture, cross-service communication, elastic scaling, and sub-second agentic latency. Mandatory for flawless deployments on Kubernetes (Local or Cloud).

install

source · Clone the upstream repo

git clone https://github.com/diegosouzapw/awesome-omni-skill

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/diegosouzapw/awesome-omni-skill "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/devops/cloud-native-engineer" ~/.claude/skills/diegosouzapw-awesome-omni-skill-cloud-native-engineer && rm -rf "$T"

manifest: skills/devops/cloud-native-engineer/SKILL.md

source content

Cloud Native Engineer (Master Skill)

This skill transforms Claude into an elite Cloud Native Architect capable of delivering production-ready distributed systems.

Core Capabilities

Strategic Domain Decomposition: Logic for splitting any monolith into clear microservice boundaries.
Standardized Dapr Infrastructure: Reliable, ready-to-use configurations for Pub/Sub, State, and Jobs.
Flawless K8s Orchestration: Deterministic deployment workflows that avoid DNS, Auth, and Probe failures.
Agentic Performance (Sub-Second): Engineering patterns for ultra-fast AI interactions.

Workflows and Procedures

1. Architectural Planning

Analyze service boundaries and define the shared communication backbone.

Guidance: MICROSERVICES.md (Domain splitting & Payload standards).

2. Infrastructure & Containerization

Build elite, secure, and lean images before rolling out the backbone.

Docker Best Practices: DOCKER_BEST_PRACTICES.md.
Backbone Setup: DAPR_CONFIG.md (Redis-first Pub/Sub and State yaml).

3. Fail-Proof Deployment (Mandatory Sequence)

Follow the EXACT order of operations to ensure 100% success.

Deployment Guide: DEPLOYMENT_GUIDE.md.
Manifest Standards: MANIFEST_STANDARDS.md.

4. Automated Orchestration

Use the "Low Freedom" scripts to automate repetitive tasks.

Deploy Minikube: scripts/deploy_minikube.ps1.
Verify Cluster: scripts/verify_cluster.ps1.

5. Diagnostic Excellence

Diagnose "Silent Failures" using the troubleshooting matrix.

Matrix: TROUBLESHOOTING.md.

6. AI Acceleration (SaaS Specialty)

Implement the persistent MCP pattern for sub-second chatbot responses.

Reference: LATENCY_OPTIMIZATION.md.

Elite Checklist for Agents

Network-Parity: Internal calls use K8s service names (
```
http://service:port
```
).
Probe-Resilience: Liveness probes have enough delay for sidecar startup.
Cluster-Auth:
```
JWKS_URL
```
points to the internal ingress/service.
Warm-Start: AI Tools are pre-initialized in the application lifespan.

AI Toolbox & Standards

To ensure maximum "intelligence" and deterministic outcomes, use the following tools and patterns:

1. Diagnostic Tools

Tool	Description	Parameters	When to Use
`kubectl logs`	Retrieve backend/frontend logs.	`-n <namespace> -c <container> --tail=<N>`	For 500 errors or startup failures.
`kubectl exec`	Run commands inside a pod.	`-n <namespace> -c <container> -- <command>`	For database checks, file verification.
`kubectl describe`	Detailed status of a resource.	`pod <name> -n <namespace>`	For `CrashLoopBackOff` or pending pods.

2. Implementation Patterns

Cascade Deletes: Always use
```
sa_column_kwargs={"ondelete": "CASCADE"}
```
for foreign keys referencing parent entities (e.g.,
```
task_id
```
in
```
TaskTag
```
).
Path Resilience: Use absolute paths (
```
/app/...
```
) inside containers to avoid "No such file" errors.
Dapr-First: Use
```
event_publisher
```
for cross-service events to maintain decoupled architecture.
SQLModel Standards: Use
```
Session(engine)
```
context managers for reliable database transactions and commits.

3. Verification Protocol

Logs: Check backend logs for tracebacks.
Reproduction: Create a minimal python script inside the pod to isolate DB/Logic errors.
Fix & Verify: Apply the fix, re-run the reproduction script, and verify via frontend/CLI.