The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because user-provided inputs are used to generate infrastructure-as-code files. * Ingestion points: Parameters like name and description defined in SKILL.md. * Boundary markers: Absent. No markers prevent user input from breaking out of HCL string literals in references/templates.md. * Capability inventory: Executes git, tofu, and gh to push changes and create Pull Requests. * Sanitization: Absent. The skill does not validate or escape user input.
[COMMAND_EXECUTION]: The skill performs automated shell commands to manage Git repositories and OpenTofu configurations. * Evidence: SKILL.md contains a sequence of commands including git checkout, tofu fmt, git push, and gh pr create.
[EXTERNAL_DOWNLOADS]: The skill fetches infrastructure configuration from a remote repository. * Evidence: Clones git@github.com:OlechowskiMichal/github-infrastructure.git in SKILL.md. This is an author-owned resource.

git-repo