file-test-bug
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local test reports and agent execution metadata files. If these files contain adversarial instructions, they could influence the agent's diagnosis or issue creation process. 1. Ingestion points:
tests/reports/junit.xml,agent-metadata.md, and test source files. 2. Boundary markers: No specific delimiters or isolation warnings are used. 3. Capability inventory: Usesgithub-mcp-server-create_issueto publish content externally. 4. Sanitization: No filtering or escaping is performed on the log content before inclusion in the issue body. - [DATA_EXFILTRATION]: The skill reads local file contents and transmits them to a repository on GitHub. Because the destination is owned by a trusted organization (Microsoft) and the activity is necessary for the bug-reporting workflow, this is a standard behavior rather than a malicious one.
Audit Metadata