review-pr

Warn

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes several local shell scripts (e.g., scripts/pr-review, scripts/pr) that take a user-provided <PR> variable. If the input is not strictly validated, it could lead to command injection. Furthermore, the skill uses the source command to load .local/review-context.env. This pattern executes the content of the file in the current shell environment, which is dangerous if the file is populated with unsanitized data from external PR metadata.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data from GitHub PR descriptions and diffs.\n
  • Ingestion points: PR descriptions and diff data retrieved via gh pr diff <PR>.\n
  • Boundary markers: No delimiters or isolation instructions are present to prevent the agent from following instructions embedded in the PR data.\n
  • Capability inventory: The agent has the ability to execute shell scripts, use the GitHub CLI to modify PR state (assigning users), and write to the local filesystem.\n
  • Sanitization: There is no evidence of sanitization or validation of the external content before it is used by the agent.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 23, 2026, 12:56 PM