deal-slip-scenario

Pass

Audited by Gen Agent Trust Hub on Nov 11, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's operational logic is entirely consistent with its described purpose of analyzing Salesforce pipeline data.
  • [DATA_EXFILTRATION]: No unauthorized network calls or exfiltration patterns were identified. All data retrieval is performed via the dispatch_readonly tool targeting official Salesforce GraphQL endpoints.
  • [COMMAND_EXECUTION]: The instructions include defensive measures that explicitly direct the agent to avoid using shell-based commands (e.g., cat, python3, ls) for reading temporary files, favoring the platform's internal Read tool instead.
  • [PROMPT_INJECTION]: No malicious prompt injection or behavior override markers were found. The connectivity testing instructions are legitimate functional requirements to handle stale connector status readouts.
  • [DYNAMIC_EXECUTION]: The dashboard widget generation uses a structured JSON template hydrated with queried data, which is a standard and safe methodology for user interface rendering.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 11, 2026, 06:39 PM