workbuddy-channel-setup
Warn
Audited by Gen Agent Trust Hub on Aug 1, 2026
Risk Level: MEDIUMPROMPT_INJECTIONCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill includes a 'Self-Evolution' section that explicitly instructs the agent to modify its own instructions and reference files based on runtime experiences. This represents a persistence mechanism where the agent can rewrite its core logic, potentially incorporating malicious behavior or instructions derived from untrusted external environments (like modified web UI elements or indirect injections) that will persist across sessions.
- [CREDENTIALS_UNSAFE]: The skill utilizes system utilities such as
pbpasteandxclipto programmatically read the contents of the user's clipboard to retrieve AppSecrets. While intended for automation, this practice exposes all data currently on the user's clipboard to the agent, posing a significant data exposure risk. - [COMMAND_EXECUTION]: The automation scripts for Feishu integration use
playwright-cli evalto execute JavaScript that targets internal React Fiber properties (__reactFiber) within the browser. This technique is used to bypass standard input restrictions and represents an unconventional use of dynamic code execution to manipulate web applications.
Audit Metadata