Enterprise Integration Skill
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill implementation includes network requests to Microsoft Graph ( https://graph.microsoft.com/v1.0), which is a well-known enterprise service. These operations follow standard patterns for extension-based authentication and do not interact with sensitive local file paths.
- [PROMPT_INJECTION]: The code patterns provided for handling chat commands demonstrate an indirect prompt injection surface when rendering data from external APIs.
- Ingestion points: External data is retrieved from Microsoft Graph API endpoints (e.g., calendar events and mail) within the
handleCalendarCommandlogic inSKILL.md. - Boundary markers: The provided examples do not use delimiters or instructions to ignore embedded content within the retrieved API data.
- Capability inventory: The skill uses
vscode.ChatResponseStream.markdownto render potentially attacker-controlled external content directly to the user interface. - Sanitization: The implementation patterns focus on functionality and do not explicitly include sanitization or escaping of data returned from the Graph API before it is processed by the agent or displayed.
Audit Metadata