The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple local helper binaries (e.g., "gstack-update-check", "gstack-config", "gstack-repo-mode", "gstack-slug") from the "~/.claude/skills/gstack/bin/" directory to handle platform configuration, session tracking, and telemetry recording.
[COMMAND_EXECUTION]: Uses dynamic execution via "eval" and "source" on the output of local helper scripts, which is a standard pattern for environment setup within the gstack framework.
[EXTERNAL_DOWNLOADS]: Includes an optional step to open an external informational URL ("garryslist.org") in the user's browser and performs version checks using a local update utility.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it ingests and processes content from various project files such as "fly.toml", "render.yaml", and GitHub Actions workflows to automate configuration.
Ingestion points: Reads project-level files including "CLAUDE.md", "fly.toml", "render.yaml", "vercel.json", and ".github/workflows/*.yml".
Boundary markers: Ingested data is processed without explicit delimiters or instructions to ignore potential instructions embedded within the configuration files.
Capability inventory: The skill has access to Bash execution, file read/write, and user interaction tools.
Sanitization: There is no evidence of sanitization or strict validation of the content extracted from external configuration files before it is used to inform agent actions.

setup-deploy