The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates scripts that execute local commands to facilitate setup workflows.
The generated scripts use the gh CLI to set GitHub Secrets and Variables.
The open_url function executes system-specific commands (wslview, explorer.exe, xdg-open, open) to open URLs in the user's browser.
The skill instructs the agent to set executable permissions (chmod +x) on the generated scripts.
[INDIRECT_PROMPT_INJECTION]: The skill performs analysis on untrusted local files to derive its operational logic, creating an injection surface.
Ingestion points: The agent reads files such as README, .env, .env.example, docker-compose.yml, and GitHub workflow files (.github/workflows/*) to identify necessary setup stages and variables.
Boundary markers: The instructions do not specify the use of delimiters or warnings when the agent processes these external files.
Capability inventory: The generated scripts have the capability to write to the local file system (write_env), execute browser-opening commands (open_url), and modify GitHub repository settings (set_secret, set_var).
Sanitization: The write_env function in the provided template.sh library performs shell-safe escaping of single quotes when writing values to the .env file.
[DYNAMIC_EXECUTION]: The primary purpose of the skill is the dynamic generation of executable bash scripts.
The skill uses a template (template.sh) to assemble a new script tailored to the user's project.
The template library uses dynamic variable assignment (printf -v "$key" '%s' "$input") to handle user-supplied configuration values at runtime.

wizard