The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the generation of GitLab CI/CD configuration files (.gitlab-ci.yml) based on user-supplied requirements, creating an attack surface where malicious data could influence the pipeline structure.
Ingestion points: User-provided project requirements and deployment details processed to create YAML configurations (SKILL.md).
Boundary markers: The instructions do not specify delimiters or "ignore instructions" markers for the content generated from user input.
Capability inventory: Includes writing configuration files and specifying execution scripts that run in automated environments.
Sanitization: No explicit sanitization or escaping of user-provided values is described for the interpolation process.
[PRIVILEGE_ESCALATION]: The skill documents the use of elevated privileges to perform system-level tasks.
Evidence: The guide includes the command sudo gitlab-runner register to configure a runner on a host system, which requires administrative access.
[COMMAND_EXECUTION]: The skill provides examples of various command-line utilities used for building, testing, and deploying applications.
Evidence: The documentation contains examples using npm, docker, aws, kubectl, git, and the gitlab-runner binary.

gitlab-ci