local-bot-review
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
coderabbit(aliased ascr) CLI to perform several operations: - Checks installation and versioning using
coderabbit --version. - Verifies authentication status with
coderabbit auth status. - Executes code reviews with various scope flags (e.g.,
cr review --agent -t uncommitted). These commands are standard for the tool's stated purpose. - [EXTERNAL_DOWNLOADS]: The skill provides instructions for adding a supplementary
code-reviewskill from thecoderabbitaiGitHub organization usingnpx skills add. It also suggests installing the core CLI via official package managers (npm or Homebrew) and explicitly recommends pinning immutable references (tags or commits) for better supply-chain security. - [DATA_EXFILTRATION]: The skill documentation acknowledges that code diffs are sent to the CodeRabbit API for processing. To mitigate the risk of data exposure, it mandates a preliminary check of the changes for secrets or credentials before the review is initiated.
- [PROMPT_INJECTION]: Includes explicit instructions to treat all review output as untrusted input. The agent is directed to verify findings against the actual code and is forbidden from executing commands or applying patches extracted from the bot's results without validation.
Audit Metadata