devops

What This Skill Does

The DevOps skill is a specialized agent designed to handle infrastructure as code (IaC) and cloud architecture tasks. Unlike traditional agents that might try to generate an entire cluster at once, this skill follows a strict, modular 'one component at a time' methodology. This ensures that resources like VPCs, subnets, and security groups are successfully provisioned and verified before moving on to complex deployments like Kubernetes clusters or database instances. It specializes in Terraform, Kubernetes, Docker, and full CI/CD pipeline orchestration, providing robust and reproducible infrastructure setups.

Installation

To integrate the DevOps skill into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/anton-abyzov/sw-devops

Ensure that your environment variables, such as cloud provider credentials (AWS_ACCESS_KEY, TF_VAR, etc.), are properly configured before initializing the skill.

Use Cases

• Automated Provisioning: Ideal for teams needing to spin up consistent development, staging, and production environments without manual configuration drift.
• CI/CD Pipeline Design: Use this skill to architect and script automated deployment pipelines that integrate seamlessly with GitHub Actions, GitLab CI, or Jenkins.
• Kubernetes Lifecycle Management: Expertly generate manifest files, Helm charts, and custom resource definitions for robust container orchestration.
• Infrastructure Auditing: Analyze existing Terraform scripts to identify potential security holes or optimization opportunities for cost reduction.

Example Prompts

1. "DevOps, please start by generating the Terraform configuration for a new VPC with two private subnets and one public subnet in the us-east-1 region."
2. "I need a Dockerfile for a Python FastAPI application that optimizes for small image size using multi-stage builds. Include a basic CI/CD workflow for GitHub Actions."
3. "Now that the VPC is ready, guide me through setting up an EKS cluster with managed node groups, ensuring the security groups are limited to my specific IP."

Tips & Limitations

To maximize the efficiency of this skill, always provide specific provider details (AWS, Azure, GCP) and region constraints early in the conversation. The 'one component at a time' approach is intentional to prevent cascading failures; please do not ask the agent to override this workflow. If a deployment fails, provide the CLI error output directly to the agent for instant debugging and remediation. Note that while the agent can generate code, it does not have direct access to your production cloud account unless you explicitly provide scoped credentials via the environment.