The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The bootstrap documentation in references/BOOTSTRAP-WORKFLOW.md provides commands to programmatically retrieve and decode the initial administrative password for ArgoCD from a Kubernetes secret. This exposes sensitive credentials in the agent's execution context.
[EXTERNAL_DOWNLOADS]: The bootstrap scripts in references/BOOTSTRAP-WORKFLOW.md download and apply Kubernetes manifests directly from the Argo Project GitHub repository. This involves the runtime execution of configuration from a remote source.
[COMMAND_EXECUTION]: The skill involves the extensive use of powerful administrative CLI tools including kubectl, git, argocd, and flux to perform cluster mutations and repository management tasks.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the ingestion and processing of untrusted user input for manifest modifications.
Ingestion points: User-provided requests for scaling, image updates, or configuration changes processed in references/WORKFLOW-STEPS.md.
Boundary markers: Absent; there are no clear delimiters to separate user data from instructions.
Capability inventory: kubectl apply, git commit, argocd sync, and flux reconcile found in SKILL.md and referenced documentation.
Sanitization: Absent; the skill lacks automated validation or sanitization of user-provided values before they are applied to Kubernetes manifests.

gitops-apply