The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's core functionality of reading and analyzing all repository files creates an attack surface for indirect prompt injection. * Ingestion points: SKILL.md instructs the agent to 'visit every single code and test file 1 at the time' and read their content. * Boundary markers: There are no instructions to use delimiters or ignore embedded instructions within the files being reviewed. * Capability inventory: The skill has the authority to write files (docs/reviews/), execute shell commands (git status, git commit), and directly modify source code ('ACTUALLY JUST FIX IT RIGHT AWAY!'). * Sanitization: The instructions do not define any filtering or sanitization process for the content ingested from the repository files.
[COMMAND_EXECUTION]: The skill uses local git commands to manage the state of the codebase during the review process. * Evidence: SKILL.md directs the agent to run git status and git commit before starting the review and after making significant changes.

full-code-review