splunk-product-question-navigator
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill ingests untrusted data from the web, which is a potential surface for indirect prompt injection. However, it implements strong defense-in-depth measures by instructing the agent to treat all retrieved content as untrusted data and ignore embedded instructions.
- Ingestion points: Public web content via the web tool as described in references/public-research.md.
- Boundary markers: Explicit directives in SKILL.md to ignore instructions found on external pages.
- Capability inventory: Access is restricted to the web tool; shell access and authenticated mutations are strictly forbidden.
- Sanitization: The response contract in references/response-and-routing.md requires paraphrasing and structured citations.
- [COMMAND_EXECUTION]: The skill includes explicit prohibitions in SKILL.md against executing commands, mutating configurations, or accessing private search results, which effectively mitigates the risk of unauthorized system actions.
Audit Metadata