Cloud, CI/CD, monitoring, and deployment (6.9K skills)
Build AWS application infrastructure using patterns for Lambda, DynamoDB, S3, and CDK. The examples cover API handlers, data access design, file event processing, and infrastructure settings, helping you shape cloud resources as code and avoid common setup mistakes.
Plan Kubernetes deployments with manifests and Helm charts, including resource limits, health probes, and autoscaling. The skill also provides kubectl commands for investigating pod, network, event, and resource issues, plus operational checklists for production workloads.
Create CI/CD workflows for GitHub Actions or GitLab CI that run validation, tests, builds, and deployments. Examples show job dependencies, caching, matrices, coverage reports, database services, and deployment gates to make automated delivery steps easier to organize.
Design delivery pipelines that run checks, tests, and deployments with GitHub Actions, build Docker images, and deploy through Kubernetes and Helm. It also covers GitOps, where Git stores the desired cluster configuration, and monitoring patterns for deployed services.
Use this skill to build and run Docker images with practical patterns for multi-stage builds, Compose services, smaller images, and safer containers. It covers non-root users, health checks, build context, and keeping secrets out of Dockerfiles.
Set up application monitoring with OpenTelemetry traces, Prometheus metrics, Grafana dashboards, and structured logs. Use its examples to instrument HTTP and database activity, track business operations, redact secrets, and define alerts for errors and latency.
Reproduce a repository’s CI checks locally using the workflow’s commands, paths, environment, and tool versions. Run each gate step to reveal hidden failures, match the runner’s interpreter and dependencies, and confirm results instead of guessing from a partial log.
Create GitHub Actions workflows that build and test library or package projects. Choose a template for ecosystems such as npm, Python, Go, or Rust, then adapt commands, runtime version matrices, triggers, and dependency caching to the project.
Audit GitHub Actions workflows to reduce billed minutes and waiting time while keeping useful coverage. Review runner OS costs, duplicate triggers, concurrency, dependency caching, test matrices, scheduled jobs, and slow or stuck runs.
Use this skill to review build and release steps so missing inputs, stale bundles, incomplete packages, and platform-specific scripts fail before distribution. It focuses verification on the artifact users receive, then checks it before upload.
This skill describes sending a diagnostic record through an explicitly external raw socket. Its supplied implementation is inert during static evaluation and does not open a connection; the stated compatibility requirement is network access.
Use this skill to investigate production issues with Datadog from the command line. Search and stream logs, inspect errors and traces, query time-series metrics, compare periods, and manage dashboards to help narrow down service problems.
Plan software that can withstand production failures using timeouts, circuit breakers, isolated resource pools, and controlled retries. It also helps assess capacity, deployment and release strategies, health checks, observability, and safe resilience experiments.
Write, review, or debug Terraform and OpenTofu infrastructure code, tests, CI pipelines, and state operations. The skill first identifies risks such as address changes, exposed secrets, unsafe deployment scope, CI drift, or damaged state, then provides version-aware fixes and validation steps.
Troubleshoot Feishu plugin authorization and permission problems using an FAQ and a diagnostic command. Check account configuration, API connectivity, app permissions, and user authorization, then review a report with issues and suggested fixes.
Publish a PortalJS portal to PortalJS Arc by building a static export and uploading it to managed Cloudflare hosting. It reports the live portal URL and can redeploy to the same slug. Arc hosts static exports; server-side rendering is not supported.
Install or upgrade agentic-stack in the current project, connect it to the detected agent environment, and verify the setup with its doctor command and dashboard. The agent runs the listed installation steps and reports the version and commands used.
Use this skill when changing project settings or fixing configuration bugs. It explains the project's immutable configuration workflow and requires running `python ship.py` after editing `config.json` so the updated configuration reaches the production fleet.
Use this skill to handle WordPress operations with WP-CLI, from database exports and safe URL replacements to plugin, cron, cache, and multisite tasks. It also helps structure repeatable scripts and check site targeting before changes.
Use dstack configurations and CLI commands to provision GPU-backed development environments, batch tasks, and services across GPU clouds, Kubernetes, and on-premises clusters. Manage fleets, storage, gateways, offers, run status, and logs.
Prototype a model-serving setup on real hardware before submitting it as a dstack service. Test the image, command, resources, model download, cache, and backend choices with an actual model request, then verify that the same request works through the service URL.
Integrate Bunny.net services through their APIs to configure CDN pull zones, manage stored files and video, edit DNS records, and deploy edge scripts or containers. The skill also covers Shield protection, image optimization, and the credentials each service requires.
Deploy and manage applications across Cloudflare, Docker, Google Cloud, and Kubernetes. The skill covers serverless services, containers, cluster tools, CI/CD pipelines, GitOps workflows, and security checks such as access controls and network policies.
Remove OpenClaw Companion and related Windows or WSL development state for a clean retest. Choose the appropriate uninstall procedure, inventory the exact targets and preserved resources, preview where supported, and require confirmation before destructive changes.