committee

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: CRITICALCOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • COMMAND_EXECUTION (MEDIUM): The skill executes a shell command to run python3 scripts/fetch_market_data.py. Executing local scripts allows the agent to perform actions on the host system, which could be malicious if the script is unverified.
  • DATA_EXFILTRATION (MEDIUM): The skill reads the Profile.md file to determine user risk preferences. This file was flagged as 'Blacklist' by a security scanner, suggesting it contains harmful links or content intended for data exfiltration.
  • EXTERNAL_DOWNLOADS (LOW): The fetch_market_data.py script is used to retrieve technical market data from external sources. This dependency on external data providers introduces a potential risk of ingesting malicious content.
  • PROMPT_INJECTION (LOW): The skill is vulnerable to indirect prompt injection because it ingests untrusted market data and processes it across multiple model rounds without explicit sanitization or boundary markers. Ingestion Point: fetch_market_data.py. Capabilities: Bash execution and file writing.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Feb 17, 2026, 05:50 PM