finance-stream-fetch
Pass
Audited by Gen Agent Trust Hub on May 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The entry point script
stream-fetch.mjsusesnode:child_processto spawnrequest.js, passing the user's financial prompt as a command-line argument. - [DATA_EXFILTRATION]: The skill transmits user-provided financial questions and authentication tokens to external servers at
aliceexp.wind.com.cn. This network activity is consistent with the skill's primary purpose of fetching remote data. - [CREDENTIALS_UNSAFE]: The skill manages API keys by reading from environment variables and configuration files located in the user's home directory (
~/.wind-aimarket/config). These credentials are sent to the API endpoint over an unencrypted HTTP connection, which is a security risk. - [PROMPT_INJECTION]: The skill takes user input directly from the
--promptargument and interpolates it into a JSON-RPC request body. There are no boundary markers or sanitization routines present, creating a surface for indirect prompt injection if the remote service is an AI agent.
Audit Metadata