crypto-market-neutral

Pass

Audited by Gen Agent Trust Hub on Apr 25, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted CSV data which constitutes an indirect prompt injection surface.\n
  • Ingestion points: The scripts/crypto_market_neutral_diagnostics.py script reads from a user-provided CSV file.\n
  • Boundary markers: There are no explicit instructions or delimiters used to warn the agent about potentially malicious instructions embedded in the CSV data.\n
  • Capability inventory: The skill performs file system read and write operations via the provided Python script and provides the summarized results to the agent context.\n
  • Sanitization: While the script converts data to numeric and datetime formats, it does not sanitize the text content against adversarial instructions intended to influence the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 25, 2026, 06:11 AM
Security Audit — agent-trust-hub — crypto-market-neutral