binance-us-asset-research
Warn
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [METADATA_POISONING]: The skill's metadata field 'author' is set to 'Binance.US', which does not match the actual skill author 'cmeiliu'. This impersonation of a well-known financial service is deceptive and could lead users to misjudge the skill's official status and safety.
- [INDIRECT_PROMPT_INJECTION]: The skill accepts user input for cryptocurrency assets and watchlists that are directly interpolated into shell command arguments.
- Ingestion points: User-provided asset symbols (e.g., BTC, SOL) passed via the
--assetand--watchlistcommand-line parameters in SKILL.md. - Boundary markers: None; asset names are used directly in the shell command without delimiters or 'ignore' instructions.
- Capability inventory: Execution of a local Python script
scripts/binance_us_brief.pywith access to the system shell. - Sanitization: No validation or escaping of user input is described in the instruction files, creating a surface for command injection if the underlying agent platform does not sanitize these arguments.
- [COMMAND_EXECUTION]: The skill relies on the execution of a Python script
scripts/binance_us_brief.py. As the script content was not included in the provided files for auditing, its internal security properties and how it handles user-provided arguments cannot be verified.
Audit Metadata