The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill determines the content of the generated bash wizard by reading repository files such as .env.example, README.md, and GitHub workflow files. A malicious repository could include deceptive content designed to trick the agent into creating a wizard that requests unnecessary secrets or includes misleading steps.
Ingestion points: Local repository files including .env, README, and .github/workflows/* as defined in SKILL.md.
Boundary markers: None present. The skill does not implement delimiters or specific instructions to ignore embedded instructions within the analyzed files.
Capability inventory: The generated scripts can write to local .env files and interact with the GitHub CLI (gh) to manage repository secrets.
Sanitization: No explicit sanitization or validation of the ingested repository content is performed before it is used to author script stages.
[DYNAMIC_EXECUTION]: The skill generates a custom bash script by populating assets/template.sh with logic derived from its analysis. While the script is intended for human execution, the generation of executable code based on potentially untrusted repository data is a notable behavior.
[COMMAND_EXECUTION]: The provided bash template utilizes standard shell commands (grep, mktemp) and the GitHub CLI (gh) to automate environment configuration. These operations are transparent and consistent with the skill's documented purpose of managing project credentials.

walk-through