pr-description-writer

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill is highly vulnerable to indirect prompt injection via the GitHub CLI.
  • Ingestion points: The skill automatically executes gh pr view --json body to read the existing description of a Pull Request. This content is attacker-controlled and considered untrusted data.
  • Boundary markers: There are no instructions to use delimiters or to ignore instructions embedded within the ingested PR body.
  • Capability inventory: The skill possesses the capability to modify repository state via gh pr edit, which includes updating the title and body.
  • Sanitization: No sanitization or validation of the ingested content is performed before the agent 'notes' it for use in generating the new PR content.
  • [Command Execution] (HIGH): The instructions for updating a PR are vulnerable to command injection.
  • Evidence: The skill instructs the agent to run gh pr edit <number> --title "<new title>" --body "...". The <new title> is wrapped in double quotes. If the agent-generated title contains shell metacharacters (e.g., $(...), `...`, or " && command), it could lead to arbitrary command execution in the shell environment where the agent operates. While the body uses a heredoc (<<'EOF'), the title parameter remains an injection vector.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 04:16 AM