The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes information from external Kubernetes clusters and Helm releases that could contain untrusted data.
Ingestion points: The skill documents commands such as helm get values, helm list, and helm status in SKILL.md to retrieve data from the environment.
Boundary markers: There are no instructions in SKILL.md to use delimiters or ignore potentially malicious instructions embedded within the output of cluster management tools.
Capability inventory: As defined in the SKILL.md frontmatter, the skill has access to Bash (restricted to helm and kubectl), Write, Edit, Glob, and Grep tools.
Sanitization: The skill does not provide methods for sanitizing or validating external data before the agent incorporates it into its decision-making context.
[EXTERNAL_DOWNLOADS]: Fetches resources and chart metadata from well-known community repositories.
Evidence: SKILL.md includes instructions for adding the Bitnami repository (https://charts.bitnami.com/bitnami) and the official Helm stable repository (https://charts.helm.sh/stable).

helm-expert