zach-search-term-report-analyzer
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches product listing information from Amazon using the curl utility to provide context for the analysis. This operation targets a well-known service (Amazon) and is essential for the skill's primary functionality.
- [COMMAND_EXECUTION]: The script scripts/browser_utils.py executes curl using the subprocess.run function. The command is invoked using a list of arguments and does not use a shell, which mitigates the risk of shell-based command injection. The target URL is constructed using a hardcoded HTTPS prefix.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from user-supplied search term reports (CSV/XLSX), creating a potential surface for indirect prompt injection where malicious data could attempt to influence the AI agent's classification logic.
- Ingestion points: Data enters the agent's context through scripts/prepare_search_term_analysis.py, which reads and aggregates user-provided report files.
- Boundary markers: The data is presented to the AI via markdown tables in roots_for_review.md for classification purposes.
- Capability inventory: The AI uses the Read tool to examine the processed data and the Write tool to save classification results, which are subsequently processed by another script to generate final reports.
- Sanitization: Standard text normalization and cleaning are implemented in scripts/clean_search_term_report.py to strip hidden characters and normalize whitespace.
Audit Metadata