The flagged credential references belong to sandbox protections and explicit key-rotation reporting, not credential exfiltration. The prompt hook sends user prompt text to Great Arrow only when installed and configured, while telemetry requires consent.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
CLI commands run on user request; the prompt suggestion hook is opt-in.
Impact
No automatic transfer of unrelated credentials or concealed payload execution was identified.
Mechanism
The hook sends the prompt and Great Arrow token to the configured service. Key rotation reports environment variable names and file paths during explicit enrollment.
Evidence
~/.great-arrow/hook-config.json
Network endpoints3
www.greatarrow.aiGAD_BASE_URLGAD_API_URL