lark-okr
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external OKR content, comments, and progress updates which could potentially contain malicious instructions.
- Ingestion points: OKR data is retrieved through commands like
+cycle-detail(references/lark-okr-cycle-detail.md),+comment-list(references/lark-okr-comment-list.md), and+progress-list(references/lark-okr-progress-list.md). - Boundary markers: Data is processed as distinct fields within the Lark API's structured JSON response, separating metadata from user-generated content.
- Capability inventory: The skill has the ability to create, update, and delete OKR records via
lark-cliacross multiple scripts. - Sanitization: Content is processed according to the vendor's
ContentBlockschema defined in references/lark-okr-contentblock.md. - [SAFE]: The skill implements safety controls for high-risk operations, such as requiring an explicit
--yesconfirmation flag for deleting data or dropping drafts (references/lark-okr-comment-delete.md). - [SAFE]: Image upload functionality is restricted to relative file paths in references/lark-okr-image-upload.md, preventing unauthorized access to system files.
Audit Metadata