business-impact-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, such as customer contracts and supplier service level agreements, which creates a surface for indirect prompt injection.
- Ingestion points: Processes service catalogues, contracts, architecture overviews, and backup records from the local environment.
- Boundary markers: No specific boundary markers or delimiters are defined for the ingested content.
- Capability inventory: Writing to the local filesystem and communicating with an external MCP server.
- Sanitization: The instructions include a directive to not include sensitive file content in external queries, but no technical sanitization is applied to the ingested data.
- [EXTERNAL_DOWNLOADS]: The skill utilizes a remote Model Context Protocol (MCP) server (https://account.ismscopilot.com/v1/account/mcp) to obtain information on compliance frameworks like DORA and NIS 2. This network dependency is a vendor-provided resource used to support the skill's analytical functions.
- [DATA_EXFILTRATION]: The skill gathers sensitive financial and operational information to calculate outage costs. While it communicates with an external vendor-controlled endpoint, the instructions explicitly restrict the agent from transmitting raw file contents or personal identifiers, mitigating the risk of unintentional data exposure.
Audit Metadata