analyzing-options-flow

Warn

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill requires access to sensitive files containing API keys and credentials for market data providers.\n
  • Evidence: Both SKILL.md and references/implementation.md direct the agent to 'Use Read tool to load API credentials from {baseDir}/config/crypto-apis.env' to authenticate and test connectivity.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data from external blockchain and market APIs.\n
  • Ingestion points: Market data from CoinGecko, Etherscan, and blockchain transactions queried via the Bash(crypto:options-*) tool as described in SKILL.md and references/implementation.md.\n
  • Boundary markers: There are no specific instructions or delimiters provided to ensure the agent ignores embedded commands or instructions within the external data it retrieves.\n
  • Capability inventory: The skill utilizes Bash(crypto:options-*), Write, and Edit tools, which could be used maliciously if the agent is influenced by injected instructions.\n
  • Sanitization: No sanitization or validation protocols are defined for the fetched external content before the agent generates reports or executes further commands.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 12, 2026, 02:53 AM