The Agent Skills Directory

[SAFE]: No malicious patterns were detected. The skill uses a professional 'draft-first' security model, meaning all emails are initially saved as drafts for human review before they can be sent. This effectively mitigates risks associated with automated or unauthorized communications.
[PROMPT_INJECTION]: The skill includes advanced defensive instructions intended to protect the agent from following malicious commands embedded in email bodies (Indirect Prompt Injection). It explicitly lists common injection patterns as examples of content to be ignored, which is a security best practice for email agents.
[COMMAND_EXECUTION]: The skill relies on the legitimate lark-cli utility to interact with the service. There are no patterns suggesting arbitrary code execution, and all sensitive operations require explicit user confirmation via a 'Step 1 (Draft) -> Step 2 (Confirm) -> Step 3 (Send)' workflow.
[DATA_EXFILTRATION]: Access to email data is restricted to the user's account following standard authentication procedures. No hidden exfiltration vectors, hardcoded credentials, or unauthorized network operations were found.

lark-mail