code-problem-analyzer
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface. It is designed to ingest and analyze untrusted external data, specifically code snippets, execution logs, and error messages provided by the user.
- Ingestion points: Untrusted data enters the agent's context during 'Step 1: 理解问题' and 'Step 2: 代码探索' via user input and file reading tools.
- Boundary markers: There are no explicit instructions to use delimiters or specific safety warnings to ignore instructions that might be embedded within the analyzed logs or code.
- Capability inventory: The skill uses
grepandreadtools to inspect files and aquestiontool to solicit further information, which could be influenced by injected instructions. - Sanitization: The analysis flow does not include steps for sanitizing or escaping the content of processed data before the LLM evaluates it.
Audit Metadata