cli-reference

Warn

Audited by Snyk on Feb 15, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (high risk: 0.80). The CLI explicitly accepts piped or command-output input (e.g., "gh pr diff | claude -p 'Review for security'") and supports reading files/outputs, which can include untrusted, user-generated third-party content (public GitHub PRs or other web-sourced data) that the agent is expected to read and interpret, creating an indirect prompt-injection risk.

MEDIUM W013: Attempt to modify system services in skill instructions.

  • Attempt to modify system services in skill instructions detected (high risk: 0.70). The prompt explicitly promotes bypassing permission prompts (e.g., --dangerously-skip-permissions), enables automated shell tooling (Bash tool, plugin-dir, --agents with tool lists) and headless CI patterns that encourage running commands without interactive permission checks, which could allow modification of machine state.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 15, 2026, 09:04 PM