oulang-data-query
Fail
Audited by Gen Agent Trust Hub on Apr 14, 2026
Risk Level: HIGHDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill directs the agent to route database operations and PostHog API requests through an external proxy at
https://cors.trigox.workers.dev/. This allows the proxy to intercept sensitive parameters including database connection strings and API tokens. \n- [DATA_EXFILTRATION]: The skill processes personally identifiable information (PII) such as user emails, phone numbers, and WeChat IDs, and transmits this data through the external proxy service. \n- [COMMAND_EXECUTION]: The instructions includecurlcommand templates designed to transmit secret credentials from agent memory and raw SQL queries to an external endpoint. \n- [PROMPT_INJECTION]: The skill lacks security measures to prevent indirect prompt injection when processing data from external sources. \n - Ingestion points: SQL query results from MySQL (via
EXTERNAL_QUERY), BigQuery tables, and PostHog event data. \n - Boundary markers: No delimiters or specific instructions are provided to the agent to distinguish retrieved data from system instructions. \n
- Capability inventory: Shell execution via
curland Python execution for authentication and data processing. \n - Sanitization: No sanitization or validation logic is present to handle potentially malicious content within the queried data.
Recommendations
- AI detected serious security threats
Audit Metadata