The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The launcher script is designed to download the redis-operator library from the author's GitHub repository at a specific pinned commit.
Evidence: The green script contains the URL https://github.com/getcolors/redis-operator.git and a pinned SHA f163b488eebbf11761b8d7f94a1507bd2cd17cad used for dependency resolution.
[DYNAMIC_EXECUTION]: The tool uses Babashka's dynamic dependency management and resolution to load its CLI and workflow logic at runtime.
Evidence: The script calls babashka.deps/add-deps to include the downloaded library and requiring-resolve to execute commands from that library.
[COMMAND_EXECUTION]: The skill executes lifecycle commands that interact with Kubernetes clusters (via kubectl context) and the DigitalOcean API.
Evidence: Usage patterns in SKILL.md include build, create, check, and drill, the latter of which involves destructive infrastructure operations (deleting Droplets) for recovery testing as a documented feature.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided configuration from colors.yml, which represents an attack surface for indirect instructions.
Ingestion points: The green script resolves and parses colors.yml to determine deployment state.
Boundary markers: Absent from the prompt instructions.
Capability inventory: The skill has high-privilege capabilities including Kubernetes resource management and cloud infrastructure modification.
Sanitization: The references/configuration.md file notes that configurations are validated twice (by the launcher and the pinned Redis package) for valid shapes, providers, and restricted SSH source IP ranges.

package-redis-operator-green