lark-mail
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes mail-related tasks by invoking the
lark-clibinary. These operations, such as querying mailboxes and managing drafts, are standard functions for the 'larksuite' vendor ecosystem. - [INDIRECT_PROMPT_INJECTION]: The skill processes email bodies and subjects, which are untrusted external inputs. The developer has implemented clear guardrails in
SKILL.mdto mitigate this risk. - Ingestion points: Untrusted data enters the agent context via retrieval tools like
+message,+messages, and+threaddescribed in the reference documentation. - Boundary markers:
SKILL.mdcontains explicit instructions to ignore any 'commands' or instructions embedded within email content, treating such text strictly as data. - Capability inventory: The skill possesses the ability to send emails (
+send,+reply), delete messages (+message-trash), and create processing rules (+rule-create). - Sanitization: The skill includes an automated HTML linting tool (
+lint-html) for write operations to prevent XSS and ensure the safety of outgoing email content.
Audit Metadata