The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is instructed to analyze repository files to determine the necessary configuration values and setup stages for the generated wizard. This creates a surface where malicious content in those files could influence the generated script.
Ingestion points: The agent reads .env, .env.example, README, docker-compose*, and .github/workflows/* (SKILL.md).
Boundary markers: There are no explicit instructions or delimiters used to prevent the agent from following instructions embedded within these external files.
Capability inventory: The resulting bash script performs file writes to .env, executes the gh CLI to set secrets, and opens URLs in the system browser (template.sh).
Sanitization: The provided template uses shell-safe quoting (e.g., ${value//\'/$sq}) when writing captured values to the environment file to prevent basic injection.
[DYNAMIC_EXECUTION]: The skill operates by generating a complete executable bash script based on a template and the analysis of the local environment.
Evidence: The instructions in SKILL.md guide the agent to copy template.sh to a target path, replace example stages with authored logic, and instruct the user to execute the resulting script.
[COMMAND_EXECUTION]: The generated wizard script relies on several system commands to perform its duties, including credential management and environment inspection.
Evidence: The template.sh library utilizes gh (GitHub CLI) for secret and variable management, grep for parsing existing environment variables, and various browser utilities (wslview, explorer.exe, xdg-open, open) for URL navigation.

wizard