The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates Bash scripts that possess capabilities to write to the local filesystem (updating .env files) and interact with the GitHub CLI (gh) to set repository secrets and variables. While these are intended functions, they involve shell command execution and credential manipulation.
[INDIRECT_PROMPT_INJECTION]: The skill determines the requirements for the generated script by scanning repository files such as README.md, .env.example, and GitHub Action workflows. This creates a vulnerability surface where malicious instructions embedded in those files could influence the generated code.
Ingestion points: Scanning of .env, .env.example, README, docker-compose*, and .github/workflows/* to identify necessary variables and secrets.
Boundary markers: The instructions do not define specific delimiters or instructions for the agent to ignore potentially malicious content within these files.
Capability inventory: The resulting script can perform file writes (write_env), open URLs in the user's browser, and execute gh commands.
Sanitization: No explicit sanitization or validation of the content read from the repository files is mentioned before it is used to scope the wizard's logic.
[DYNAMIC_EXECUTION]: The core purpose of the skill is to generate executable scripts from a template (template.sh). This is categorized as dynamic code generation. Although the instructions recommend that the user review the script and perform a static trace, the automated creation of executable code based on untrusted repository data carries inherent risk.

wizard