The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of code review feedback and PR comments.
Ingestion points: The skill ingests 'PR comments' and 'verbal review notes' as core inputs (SKILL.md).
Boundary markers: There are no defined technical delimiters or 'ignore' instructions for the external content, though the agent is instructed to verify claims against the codebase.
Capability inventory: The skill utilizes git for repository inspection and gh for posting API responses.
Sanitization: No specific escaping or sanitization of the review feedback is performed before it is processed or used in tool arguments.
[COMMAND_EXECUTION]: The skill utilizes development tools to interact with repositories and pull requests.
Uses git to verify code-related claims.
Uses the gh api command to post replies back to specific comment IDs on GitHub.
[METADATA_POISONING]: The skill displays conflicting metadata information.
The README.md file explicitly states that the 'Ship Shit Dev' attribution was incorrect and has been removed.
The plugin.json file still lists 'Ship Shit Dev' as the author and provides corresponding contact URLs.

receiving-code-review