qwen-runner

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/run_qwen.py wraps the qwen CLI using subprocess.run() with argument lists. This design correctly avoids shell injection vulnerabilities by ensuring that user-provided input is not interpreted as shell commands.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests content from local files and interpolates it into prompts for an LLM.
  • Ingestion points: Files are read via --prompt-file, --session-file, and --output-schema arguments in scripts/run_qwen.py.
  • Boundary markers: The script prefixes sections with labels like 'Role:', 'Execution constraint:', and 'Prior conversation context:', providing some structure, though it does not use hardened delimiters to isolate untrusted file content.
  • Capability inventory: The skill has the capability to execute shell commands via the qwen CLI wrapper in scripts/run_qwen.py.
  • Sanitization: File content is read and interpolated directly into the prompt without sanitization or validation of its contents.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 02:19 PM
Security Audit — agent-trust-hub — qwen-runner