The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous templates and instructions for executing powerful command-line tools including kubectl, terraform, docker, aws, and gh. These are used for deploying infrastructure, managing containers, and performing incident forensics (e.g., tcpdump in collect-evidence.sh).
[DYNAMIC_EXECUTION]: The skill includes Python and Shell script templates for automation tasks such as auto-remediation of incidents, platform engineering APIs, and release orchestration. These scripts are intended for use by a DevOps engineer to automate operational workflows.
[INDIRECT_PROMPT_INJECTION]: As a DevOps assistant, the skill ingests external data such as application logs, CI/CD pipeline outputs, and configuration files. This creates an attack surface where malicious content in logs could attempt to influence the agent's behavior.
Ingestion points: System logs and deployment metadata are processed during incident response (references/incident-response.md) and release verification.
Boundary markers: The skill requires validation steps like terraform plan and configuration linting.
Capability inventory: Full file system access, network operations, and administrative tool execution (kubectl, terraform, docker).
Sanitization: The 'Core Workflow' in SKILL.md mandates explicit user approval before any production or customer-facing deployment, providing a critical human-in-the-loop safety check.

devops-engineer