sql-helper

Pass

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes natural language input from users to generate executable-style SQL code.
  • Ingestion points: User-provided query requirements, database schema descriptions, and error logs are ingested and processed in the 'Step 1: 理解需求' and 'Step 2: 编写/优化 SQL' phases.
  • Boundary markers: The instructions do not implement boundary markers (such as XML tags or explicit delimiters) to isolate user-provided data from the core system instructions.
  • Capability inventory: The skill's capabilities are restricted to text-based generation of SQL, DDL, and DML statements; it does not possess the capability to execute commands, access the file system, or initiate network requests.
  • Sanitization: There are no internal mechanisms to sanitize or filter user input for adversarial prompt patterns before it is used to construct the final output.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 1, 2026, 03:44 PM
Security Audit — agent-trust-hub — sql-helper