NYC

dspy

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • External Downloads (LOW): The skill instructs users to install the 'dspy-ai' package via pip. This is a well-known, legitimate package from the Stanford NLP group.
  • Indirect Prompt Injection (LOW): The skill provides a framework for ingesting untrusted data to programmatically generate prompts (e.g., in the RAGModule and QA signatures). This creates a standard attack surface where external content could influence LLM output.
  • Ingestion points: Input fields defined in 'QA' and 'Analysis' signatures, as well as the 'Retrieve' module in the RAG pattern.
  • Boundary markers: Not explicitly detailed in the summary, though the framework uses signatures to define data types.
  • Capability inventory: Prompt generation and LLM interaction via the configured language model.
  • Sanitization: No specific sanitization or escaping of input data is demonstrated in the provided examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:30 PM