dingtalk-docs-reader

SUSPICIOUS: The skill is internally coherent for reading DingTalk docs, and it does not show clear malware or third-party credential exfiltration. However, it depends on manual extraction of full browser session cookies and reverse-engineered internal endpoints to bypass the official app/admin auth model, which creates medium security risk disproportionate to a normal supported integration.

The code implements a document client and does not show clear malicious behavior. Redirect handling can leak the supplied DingTalk cookie and XSRF token if an API endpoint responds with a redirect to an attacker-controlled host. Validate redirect destinations and avoid forwarding credentials across origins.