The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The green launcher script fetches necessary automation logic from the author's verified GitHub repositories (github.com/getcolors/mysql-ha, github.com/getcolors/green, and github.com/getcolors/once). These dependencies are pinned to a specific Git commit SHA to ensure integrity and prevent supply chain attacks.
[DYNAMIC_EXECUTION]: The skill uses Babashka's add-deps function to dynamically load and resolve the GitHub-hosted libraries at runtime. This allows the launcher to maintain a small footprint while integrating complex infrastructure logic from trusted source trees.
[COMMAND_EXECUTION]: To manage the cluster, the skill orchestrates the execution of standard infrastructure tools including OpenTofu for resource provisioning, Ansible for node configuration, and SSH for secure communication with cluster members.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from a colors.yml configuration file to define cluster parameters.
Ingestion points: The green script automatically locates the colors.yml file within the project directory structure.
Boundary markers: Configuration input is limited to a flat YAML map and validated against the YAML 1.2 core schema to prevent malformed or malicious data injection.
Capability inventory: The ingested configuration drives subsequent OpenTofu and Ansible operations on the database cluster.
Sanitization: The skill enforces strict separation of concerns by ensuring that only non-secret values are stored in the YAML configuration, while all sensitive credentials must be provided through external environment variables.

package-mysql-ha-green