google-file-search

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Prompt Injection (LOW): The skill implements a Retrieval-Augmented Generation (RAG) workflow, creating an attack surface for indirect prompt injection via document content.\n
  • Ingestion points: External documents (PDF, DOCX, TXT, etc.) are read and uploaded for indexing in scripts/upload_documents.py and templates/python-client.py.\n
  • Boundary markers: The prompt construction in scripts/search_query.py (line 62) and templates/python-client.py (line 166) lacks specific delimiters (like XML tags or markdown blocks) or instructions to the model to ignore embedded instructions within retrieved grounding text.\n
  • Capability inventory: The skill uses the google-genai client to generate content and manage file stores, allowing retrieved data to influence model outputs.\n
  • Sanitization: No sanitization, filtering, or validation of document content is performed prior to the indexing or retrieval process.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:32 PM