Skills
Navigate
skills/aahl/skills/crypto-report/Gen Agent Trust Hub

crypto-report

Fail

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The script scripts/theblockbeats-news.sh connects to api.theblockbeats.news, a domain flagged by automated scanners for phishing.
  • [COMMAND_EXECUTION]: The skill executes local bash scripts (scripts/binance-ai-report.sh and scripts/theblockbeats-news.sh) to fetch and process cryptocurrency data.
  • [INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by processing and displaying untrusted external news content and market reports. Ingestion points: API responses from binance.com and theblockbeats.news processed in the shell scripts. Boundary markers: None; content is extracted and displayed without delimiters or warnings to the agent. Capability inventory: Shell script execution, network access via curl, and JSON parsing via jq. Sanitization: The skill extracts specific fields using jq, but it does not perform any text-level filtering to prevent malicious instructions within the fetched news from influencing the agent's behavior.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Sep 15, 2026, 03:26 PM
Security Audit — agent-trust-hub — crypto-report