The material explicitly states that no keys or environment variables are required, and it does not request account tokens, API keys, or other sensitive credentials, so credential exposure appears limited.

The README allows URLs inside <qqmedia> and says user-sent images are automatically downloaded locally, indicating potential network ingress/egress through media URLs or platform attachment flows. No specific remote endpoints are listed and there is no clear red flag of unrelated third-party exfiltration, but media content may still traverse QQBot/platform channels and should be scoped carefully.

Based on the material, this is a tag-based media capability description, and the objective checks also mark it as prompt-only; there is no indication of running scripts locally, installing extra binaries, or requesting elevated system privileges.

The material says received images are automatically downloaded locally and requires outbound local files to reside under ~/.openclaw/media/qqbot/ or its subdirectories, showing that it handles local media paths and attachment data. The described scope appears centered on a trusted media directory, with no explicit request to read arbitrary system locations, but attachment storage and resend behavior still warrant least-privilege review.

The source is an open-source GitHub repository with extremely strong community adoption (about 377k stars), which are clear risk-reducing signals. While the license is undeclared and maintenance status is unknown, adding some governance uncertainty, the current material shows no sign of closed-source distribution, provenance distortion, or suspicious delivery.