zach-search-term-analyzer

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill performs local data analysis on user-provided CSV files. No network operations or external data transmissions are performed during script execution.\n- [SAFE]: The dependencies listed in requirements.txt (pandas, numpy, matplotlib, seaborn, openpyxl) are standard, reputable packages from official package registries.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface where untrusted data from Amazon reports is ingested by the agent via generated markdown files. While it lacks explicit boundary markers to delimit external content, this processing is essential to the skill's primary function and no malicious exploitation patterns were observed. \n
  • Ingestion points: CSV files located in the user-specified data_dir and processed by scripts/analyze_search_terms.py.\n
  • Boundary markers: Absent. The generated markdown reports (analysis_reports/*.md) do not include explicit delimiters or warnings to the LLM to ignore instructions within the data.\n
  • Capability inventory: The skill uses Read, Glob, Bash, Write, Edit, and Grep.\n
  • Sanitization: The script does not implement specific sanitization or filtering to prevent prompt injection control characters from being included in the generated reports.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 12:35 PM
Security Audit — agent-trust-hub — zach-search-term-analyzer