The flagged credential and encryption code fits the CLI's device key-rotation feature: environment key names are reported only during explicit enrollment, which requires browser approval. The separate prompt hook is opt-in, and runtime telemetry requires consent.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
User runs the gad CLI; key reporting requires `gad keys enroll`, and prompt suggestions require opt-in hook installation.
Impact
Observed network activity supports the CLI's declared features; no malicious impact was identified.
Mechanism
The enrollment flow reports environment key names and paths to Great Arrow's key-rotation API, while the opt-in prompt hook sends the prompt to the Great Arrow service. No automatic credential-value exfiltration was found.
Evidence
dist/index.jsdist/hooks/gad-suggest-tools.mjs~/.great-arrow/hook-config.json
Network endpoints3
www.greatarrow.aiGAD_API_URL-configured service endpointGAD_BASE_URL-configured hook endpoint