The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes git, gh, and pnpm through shell commands that include user-supplied parameters like issue numbers, branch names, and PR titles.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the user and external GitHub issues, which is then interpolated into shell commands.
Ingestion points: User descriptions and GitHub issue content gathered in SKILL.md.
Boundary markers: Absent for the interpolation of variables into shell scripts.
Capability inventory: Execution of git, gh, and pnpm; modification of repository files.
Sanitization: No explicit validation or escaping is applied to user-provided strings before they are passed to CLI tools.
[DYNAMIC_EXECUTION]: The skill modifies project source code and runs tests via pnpm test, executing code that was generated or modified at runtime.

hotfix