crud

Warn

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill is instructed to automatically read ruoyi-admin/src/main/resources/application-dev.yml. In standard Spring Boot applications, this file contains sensitive plain-text credentials including database usernames, passwords, and connection URLs.
  • [COMMAND_EXECUTION]: The execution flow involves running database commands SHOW CREATE TABLE and DESC using user-supplied table names. Without strict validation of the table name input, this presents a surface for SQL-based metadata exploration or potential injection.
  • [DATA_EXFILTRATION]: The skill performs broad read operations on the local file system, including sensitive configuration files and core service/controller source code (e.g., SysNoticeController.java). This exposes internal application logic and architecture to the model context.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through several vectors:
  • Ingestion points: Reads database schema metadata, table structures, and existing Java source files.
  • Boundary markers: No explicit boundary markers or instructions to ignore embedded commands within the processed files are defined.
  • Capability inventory: Includes file system read/write access and database command execution capabilities.
  • Sanitization: No sanitization logic is specified for the data retrieved from the database or existing source files before it is processed by the model.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 7, 2026, 09:04 AM