ict-risk-framework-check
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes a variety of user-provided organizational documents (e.g., policies, audit reports, meeting minutes). This creates an attack surface where malicious instructions embedded within those documents could attempt to influence the agent's output.
- Ingestion points: Processes licences, risk framework documents, asset inventories, and management body minutes in Step 1 and Step 2.
- Boundary markers: No specific delimiters or "ignore embedded instructions" markers are defined for the content extracted from input documents.
- Capability inventory: File system write operations to
./ict-risk-framework-check/and network communication via the ISMS Copilot MCP server. - Sanitization: The skill lacks explicit instructions for sanitizing or escaping the text content extracted from user documents before processing.
- [EXTERNAL_DOWNLOADS]: The skill integrates with an external MCP server for framework lookups.
- Evidence: Mentions the MCP server endpoint
https://account.ismscopilot.com/v1/account/mcpfor the ISMS Copilot service. - Context: The skill provides clear safety instructions to the agent to avoid including secrets, file contents, or personal data in external queries, reducing the risk of accidental data exposure.
Audit Metadata