stock-trading
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes several Python scripts that perform market data retrieval and simulated trade execution. These scripts are intended to be called by the agent through the command line to manage the user's simulated portfolio.
- [INDIRECT_PROMPT_INJECTION]: The skill fetches data from external market providers like Sina Finance and the OpenCTP simulation front-ends. 1. Ingestion points: Data is ingested via HTTP requests and CTP protocol connections in the get_price.py and trade.py scripts. 2. Boundary markers: No explicit instructions are provided to the agent to treat market data as untrusted or to ignore instructions embedded in data strings. 3. Capability inventory: The agent can perform network requests and write to the local state.json file to update portfolio status. 4. Sanitization: String data such as asset names are displayed without transformation, which could be an injection vector if the data source provided malicious text.
- [DATA_EXPOSURE]: The skill manages authentication for the OpenCTP simulator by reading from OPENCTP_USER and OPENCTP_PASSWORD environment variables, which prevents the need for hardcoding sensitive information within the skill files.
Audit Metadata