anfra-aql

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill documents the use of AI functions such as ai_summarize, ai_complete, and ai_classify that process potentially untrusted data fields (e.g., reviews.content in references/examples/56.md). This creates a vulnerability surface where malicious instructions embedded in the data could influence the behavior of the AI models processing the AQL queries.
  • Ingestion points: reviews.content (documented in references/examples/56.md) and other text fields passed to AI functions described in references/aql/ai-functions.md.
  • Boundary markers: The AQL language documentation does not specify boundary markers or sanitization requirements for text data processed by AI functions.
  • Capability inventory: The skill provides knowledge on executing AQL, which includes powerful data transformation and AI-driven summarization, classification, and completion capabilities.
  • Sanitization: No sanitization or escaping of external content is documented for the AI functions.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 03:35 AM