pm-advisor-yujun
Pass
Audited by Gen Agent Trust Hub on Jul 14, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses the
WebSearchtool to gather information about product features, competitors, and user scenarios (Step 2 inSKILL.md). This process ingests untrusted external data into the agent's context, creating a potential surface for indirect prompt injection attacks. - Ingestion points: Data is fetched from the web using the
WebSearchtool and incorporated into the decision-making framework. - Boundary markers: There are no instructions or delimiters provided to the agent to differentiate untrusted search results from system instructions or to ignore embedded commands.
- Capability inventory: The agent is restricted to searching the web and generating responses; it lacks capabilities for file modification, code execution, or network exfiltration.
- Sanitization: The skill does not define any validation, filtering, or escaping mechanisms for the content retrieved from the web.
Audit Metadata