NYC

dspy-framework

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: CRITICAL
Full Analysis
  • SAFE: The provided metadata file contains no executable code, malicious instructions, or obfuscated content.
  • External Downloads (SAFE): The skill specifies installation of the legitimate dspy-ai package and standard provider libraries (OpenAI, Anthropic). No suspicious or typosquatted dependencies were identified.
  • Indirect Prompt Injection (LOW): By design, the DSPy framework optimizes prompts using external data, creating an inherent surface for indirect prompt injection.
  • Ingestion points: Training and evaluation datasets processed by modules such as BootstrapFewShot and MIPROv2.
  • Boundary markers: None specified in metadata; standard DSPy signatures typically do not enforce strict delimiters for training data.
  • Capability inventory: The framework facilitates multi-hop reasoning, agent routing, and code generation (ProgramOfThought).
  • Sanitization: No explicit sanitization or validation logic is defined in the metadata for the ingested data.
  • Automated Scan Note: The scanner alert regarding self.logger.info appears to be a false positive, as that string is a common logging call and is not present in the analyzed file.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Feb 17, 2026, 05:30 PM