walmart-review-checker
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill ingests untrusted product reviews which creates a surface for indirect prompt injection attacks. Malicious reviews could contain instructions to alter the authenticity score or manipulate the agent's behavior. * Ingestion points: Review content provided via text prompts or JSON input in SKILL.md. * Boundary markers: None identified; the skill does not use delimiters or instructions to isolate user-provided data. * Capability inventory: Executes shell commands via scripts/analyzer.py. * Sanitization: No input validation or sanitization logic is visible.
- [COMMAND_EXECUTION]: Usage examples in SKILL.md involve executing a Python script with JSON input directly in the command line. This can lead to command injection if the input reviews contain shell metacharacters or broken quoting patterns.
- [EXTERNAL_DOWNLOADS]: The skill installation instructions use npx to download a skill package from an external source (nexscope-ai/eCommerce-Skills), which was not provided for analysis and originates from an unverified repository.
- [NO_CODE]: The logic for analysis is contained in scripts/analyzer.py, which is missing from the provided file list, preventing a full security audit of the execution behavior.
Audit Metadata