The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes local repository files, such as README, .env files, and GitHub workflow configurations, to determine the necessary setup steps. These files represent an untrusted input surface that could contain malicious instructions designed to manipulate the generated script's logic.
Ingestion points: Reads from .env, .env.example, README, docker-compose.yaml, and .github/workflows/* to scope the procedure.
Boundary markers: The skill does not implement specific delimiters or 'ignore' instructions when reading these files to prevent embedded commands from influencing the agent.
Capability inventory: The generated wizard script has the capability to write to the local filesystem (.env) and update remote GitHub repository secrets or variables via the gh CLI.
Sanitization: There is no evidence of sanitization or validation performed on the data extracted from the repository files before it is used to template the generated bash script.
[DYNAMIC_EXECUTION]: The core functionality of the skill is the generation of new bash scripts (.sh). The agent is instructed to create these scripts based on a template, set them as executable (chmod +x), and provide them to the user for execution.
[COMMAND_EXECUTION]: The generated scripts are designed to execute shell commands and interact with the GitHub CLI (gh) to manage sensitive repository configuration, such as setting secrets and variables.

wizard