The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the rg (ripgrep) utility to scan the repository's file structure and identify existing configuration or documentation files.
[REMOTE_CODE_EXECUTION]: The instructions direct the agent to "finish by running the repo's validation command." This pattern allows for the execution of arbitrary shell commands or scripts defined within the repository being scaffolded, which constitutes a risk if the repository content is untrusted or malicious.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection by design, as it relies on content found within the target repository to determine which commands to execute.
Ingestion points: Repository files and structures identified by the rg command in SKILL.md.
Boundary markers: None; there are no delimiters or instructions to treat repository-sourced commands as untrusted.
Capability inventory: File system inspection via rg and arbitrary command execution via shell to run "validation" scripts.
Sanitization: The skill lacks any instructions to validate, sanitize, or confirm the safety of the validation commands before they are executed by the agent.

repo-scaffold