The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local bash script scripts/compare_default_branch.sh and various standard git commands (such as git rev-list, git diff, and git remote show) to summarize differences between branches. These operations are restricted to analyzing the local repository metadata and code state.
[PROMPT_INJECTION]: The skill processes untrusted data in the form of git diffs and commit messages from the repository being reviewed, which is a surface for indirect prompt injection.
Ingestion points: Untrusted data enters the agent context through the execution of git diff and git rev-list commands in SKILL.md.
Boundary markers: The instructions do not define explicit delimiters to separate repository content from the agent's internal reasoning or instructions.
Capability inventory: The skill has capabilities for shell command execution (via git) and reading local repository files.
Sanitization: No validation or escaping of the diff content is performed before it is analyzed by the agent, allowing potential instructions embedded in the code to be processed as part of the review workflow.

finish-review