The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using subprocess.run() in scripts/publish_static.py to perform git operations (clone, add, commit, push) and to run the image optimization script scripts/optimize_images.py. User input provided via the --slug argument is sanitized through a regex filter ([^a-z0-9]+) before being used in directory paths or commit messages, reducing the risk of command injection.
[EXTERNAL_DOWNLOADS]: The scripts/publish_static.py script clones source code from GitHub repositories using the git clone command. This operation targets a well-known service and is a core requirement for the skill's deployment functionality.
[INDIRECT_PROMPT_INJECTION]: The skill processes external static site artifacts which represent a surface for indirect prompt injection.
Ingestion points: Artifact files (HTML, CSS, JS) are read and copied from the local source directory specified in the --source argument within scripts/publish_static.py.
Boundary markers: There are no technical delimiters or instructions to ignore instructions embedded within the artifact files during processing.
Capability inventory: The skill has the capability to perform network operations (git push) and write to the local file system.
Sanitization: Content within the artifacts is not sanitized; however, the SKILL.md instructions direct the agent to manually verify the content for secrets and absolute paths before publishing.

github-pages-publisher