The Agent Skills Directory

[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes untrusted data from GitHub pull requests to perform its analysis.
Ingestion points: The skill retrieves pull request metadata (title, body) and code changes (diffs) using gh pr view and gh pr diff during the summarization and parallel reviewer steps.
Boundary markers: There are no explicit instructions to use delimiters or boundary markers to isolate the untrusted PR content from the agent's instructions.
Capability inventory: The skill can read repository files, execute shell commands via git and gh, and perform write operations by posting comments to GitHub using gh pr comment.
Sanitization: The instructions do not specify any validation or sanitization of the PR content before it is processed by the reviewer agents.
[COMMAND_EXECUTION]: The skill relies on the execution of external command-line tools to fulfill its primary purpose.
Evidence: The workflow explicitly instructs the use of git diff, gh pr view, gh pr diff, and gh pr comment to interact with the local environment and the GitHub service.

code-review