The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches content from external GitHub repositories using the gh API. This is a core part of its functionality to allow comparing local skill files with upstream versions. As GitHub is a trusted service, these downloads are considered safe within this context.
[COMMAND_EXECUTION]: The bash script uses standard system utilities including gh, diff, awk, sed, and grep. It implements security best practices by validating user-supplied skill names and repository identifiers against strict regular expressions before incorporating them into shell commands.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection attack surface.
Ingestion points: The script in scripts/compare.sh fetches SKILL.md content from remote GitHub repositories.
Boundary markers: While the output is labeled as a raw diff, there are no specific instructions or delimiters used to ensure the agent ignores any embedded instructions within that external content.
Capability inventory: The skill has the ability to execute subprocesses and interact with the GitHub API.
Sanitization: The external markdown content is displayed to the agent in its raw form without escaping or filtering. A maliciously crafted upstream file could contain instructions that influence the agent's behavior during the interpretation protocol.

skill-diff