test-plan
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-generated requirement text and acceptance criteria, creating a potential surface for indirect prompt injection attacks.\n
- Ingestion points: Data is retrieved from the
requirement_gettool via thedescriptionandacceptanceCriteriafields inSKILL.md.\n - Boundary markers: The skill contains clear instructions to the agent in
SKILL.mdto treat these fields as data rather than instructions and to explicitly ignore any commands found within them.\n - Capability inventory: The skill utilizes
testcase_createto perform write operations, as documented inSKILL.mdandreferences/case-taxonomy.md.\n - Sanitization: The skill instructions require the agent to report any suspected instructions in the source text instead of executing them.
Audit Metadata