The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes the colors.yml configuration file to drive deployment actions.
Ingestion points: colors.yml file is the primary input processed by the green launcher script.
Boundary markers: No specific delimiters or "ignore embedded instructions" markers are defined for the configuration file content.
Capability inventory: The skill executes OpenTofu, Ansible, and Docker commands; writes SSH configuration and identity files; and performs network operations via cloud provider APIs (Vultr, DigitalOcean, Cloudflare).
Sanitization: The instructions recommend kebab-case keys and non-secret values, but the underlying execution logic for parsing is external to the skill files.
[DYNAMIC_EXECUTION]: The green launcher script dynamically loads its execution logic at runtime.
Evidence: The script uses babashka.deps/add-deps to fetch and add the signoz, green, and once libraries from GitHub to the classpath.
Evidence: Functions such as green.cli/run-cli are invoked using requiring-resolve after the libraries are loaded.
[EXTERNAL_DOWNLOADS]: The skill downloads code and binaries from external sources during execution.
Evidence: The launcher script fetches library dependencies from github.com/getcolors/ repositories using specific Git SHAs.
Evidence: The histogramQuantile binary is downloaded from GitHub releases as part of the application setup, as noted in references/configuration.md.

package-signoz-green