The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell execution (denoted by the '!' prefix) to run various git commands and text processing utilities like sed, sort, and uniq to gather repository context.
[COMMAND_EXECUTION]: In 'Step 1: Determine Diff Scope', the skill identifies that it will use user-provided arguments (such as branch names or commit ranges) to construct and run a git diff command. This represents a potential command injection vector if the input is not sanitized by the underlying agent framework.
[PROMPT_INJECTION]: The skill is subject to Indirect Prompt Injection risks because it reads and analyzes the full content of changed files in 'Step 2: Build Context'. An adversary could commit code containing hidden instructions designed to manipulate the agent's review logic.
Ingestion points: The skill reads the full content of all files identified in the git diff scope.
Boundary markers: The instructions do not specify the use of delimiters or protective headers when the agent processes the file contents.
Capability inventory: The skill possesses the ability to execute Bash commands (limited to git operations), Read files, and use Grep/Glob tools.
Sanitization: No sanitization or filtering of the file content is performed before it is analyzed by the LLM.

code-review