is-this-a-problem
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data via the
[问题](Question) argument to perform pattern matching and risk classification. It lacks explicit boundary markers or sanitization steps to prevent adversarial instructions embedded in user queries from being executed as system commands. - Ingestion points: User-supplied text provided in the
[问题]argument of the/is-this-a-problemcommand (defined in SKILL.md). - Boundary markers: The skill does not instruct the agent to use delimiters or ignore potentially malicious instructions within the user's question.
- Capability inventory: The skill reads internal risk calibration files (
CLAUDE.md) and manages workspace context by reading and writing files within the~/.claude/plugins/config/claude-for-legal-zh/directory. - Sanitization: No methods for input validation, escaping, or filtering are present for the processed text.
Audit Metadata