llm-issue-discovery
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of LLM logs, traces, and responses to identify failure patterns.
- Ingestion points: The skill instructs the agent to ask for and process logs in formats like JSON, CSV, or pasted text (Step 1 in
SKILL.md). - Boundary markers: There are no explicit instructions to use delimiters or 'ignore embedded instructions' warnings when analyzing the ingested logs.
- Capability inventory: The skill's primary capabilities are text analysis and generating a structured report. It does not perform automated file system writes, network requests, or command execution based on the data.
- Sanitization: No sanitization or validation of the log content is performed prior to processing.
Audit Metadata