The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill provides a pattern for 'Dynamic Child Pipelines'. This involves a script (generate_pipeline.py) generating a YAML configuration file at runtime, which is then loaded and executed as a sub-pipeline by the GitLab runner.
[INDIRECT_PROMPT_INJECTION]: The dynamic pipeline generation pattern creates a vulnerability surface where untrusted data could influence the pipeline's execution.
Ingestion points: The generate_pipeline.py script mentioned in SKILL.md is the primary entry point for dynamic content.
Boundary markers: None are present in the provided templates to distinguish between static instructions and dynamically generated content.
Capability inventory: The triggered child pipeline has the capability to execute shell commands, access CI/CD variables (including secrets), and interact with deployment targets.
Sanitization: There is no evidence of input validation or sanitization for the data used to generate the child pipeline configuration.
[CREDENTIALS_UNSAFE]: The multi-environment deployment template includes a command to configure kubectl with --insecure-skip-tls-verify=true. This setting disables the verification of the server's SSL/TLS certificate, making the connection susceptible to Man-in-the-Middle (MitM) attacks which could lead to the exposure of the $KUBE_TOKEN or other sensitive cluster communications.
[EXTERNAL_DOWNLOADS]: The skill references several official Docker images from well-known services and organizations, including node, docker, bitnami/kubectl, hashicorp/terraform, and aquasec/trivy for use in CI/CD stages.

gitlab-ci-patterns