investigation-summary
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill is designed to process and summarize external "privileged investigation memos." If these memos contain adversarial text or instructions disguised as case data, they could potentially influence the agent's behavior or cause it to ignore the "audience-stripping" rules defined in the skill.
- Ingestion points: SKILL.md indicates the agent loads and processes external investigation memos as primary input.
- Boundary markers: The provided instructions do not specify explicit delimiters or "ignore embedded instructions" markers for the memo content.
- Capability inventory: No dangerous execution capabilities (subprocesses, file writes, or network operations) were identified within this specific skill definition.
- Sanitization: The skill lacks explicit instructions to sanitize or validate the content of the memo before summarization to ensure embedded text does not override the summary logic.
- Sensitive Data Handling: The skill manages highly sensitive information, including legal exposure analysis and attorney mental impressions. While it provides instructions for audience-specific stripping, there is a consideration regarding the reliability of the agent to consistently exclude privileged details from audiences like HR as intended by the workflow.
Audit Metadata