bjtuo-classroom-query

Warn

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/query_classroom.py uses eval() within the get_captcha_code function to calculate results for mathematical captchas. While it uses a regular expression ([\d\+\-\*\/]) to filter the input, using eval() on content derived from an external source (AI model OCR output) represents a dynamic execution risk.
  • [EXTERNAL_DOWNLOADS]: The skill makes network requests to a third-party, non-official API (http://yaya.csoci.com:2333/api/classnum/) to fetch real-time classroom occupancy data. This introduces a dependency on an unverified external service.
  • [CREDENTIALS_UNSAFE]: The execution script scripts/query_classroom.py attempts to load credentials (BJTU_USERNAME, BJTU_PASSWORD, ZHIPU_API_KEY) from .env and secrets.json files located in parent directories (../../), which may expose sensitive information if not properly restricted.
  • [DATA_EXFILTRATION]: The skill saves browser session cookies and authentication tokens to a local file auth_state.json. If this file is not secured, it could lead to session hijacking or unauthorized access to the user's university account.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 16, 2026, 07:27 AM