benchmark-topic-radar
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a standard text-based workflow that uses local reference files to guide the agent's behavior. It does not perform any sensitive operations such as network requests, file writes, or shell command execution.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied content and reference samples, which is a potential surface for indirect prompt injection, though the impact is negligible due to the lack of available tools.\n
- Ingestion points: Creator content and reference samples provided in user prompts (SKILL.md).\n
- Boundary markers: The skill instructs the agent to differentiate between signals and follow safety rules (references/reference-safety-rules.md), though it lacks specific input delimiters.\n
- Capability inventory: There are no subprocesses, network tools, or file system access capabilities present in the skill definition.\n
- Sanitization: No specific input sanitization steps are defined, relying on the agent's internal safety guardrails.
Audit Metadata