lark-cli
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill suggests using
npx -y @larksuite/clias a fallback, which downloads and executes the official command-line tool for LarkSuite from the public NPM registry. - [COMMAND_EXECUTION]: Relies on the
lark-clibinary to perform administrative and user tasks, including checking authentication status and listing available skills. - [DATA_EXFILTRATION]: Accesses sensitive corporate data from Lark/Feishu, such as chat messages, wiki documents, and multi-dimensional tables, as part of its primary functionality.
- [PROMPT_INJECTION]: The skill ingests untrusted data from external sources, specifically Lark/Feishu chat messages and documents (Ingestion points:
lark-cli imand document reading). It lacks explicit boundary markers or sanitization logic to differentiate between data and instructions. The agent environment provides capabilities for shell command execution and writing back to the platform (Capability inventory:lark-clioperations defined inSKILL.md).
Audit Metadata