filing-bug-reports
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill populates report templates using data from conversation history and external tool outputs, which constitutes a potential surface for indirect prompt injection where malicious instructions could be embedded in data. Ingestion points: Conversation history and output from the evaluate_matlab_code tool are ingested in SKILL.md. Boundary markers: The skill utilizes the structure provided in assets/bug-report-template.md but lacks specific delimiters to isolate external content from instructions. Capability inventory: The skill can execute MATLAB code via evaluate_matlab_code and perform file-write operations to the local issues/ directory. Sanitization: The skill includes a clear guardrail to redact secrets, tokens, and PII using the tag.
- [COMMAND_EXECUTION]: The skill executes diagnostic MATLAB commands (such as ver, computer, and connector port checks) via the evaluate_matlab_code tool. These operations are used exclusively to gather environment details necessary for reproduction and are consistent with the skill's intended purpose.
Audit Metadata