lark-workflow-task-digest
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Uses
lark-clito interact with the Lark platform and the systemdatecommand. These are legitimate uses for the skill's purpose. - [DATA_EXFILTRATION]: Transfers task summaries to Lark's messaging and document systems, which is the intended functional outcome.
- [PROMPT_INJECTION]: Use of emphatic terms like 'CRITICAL' in the instructions relates to workflow dependencies and does not represent a safety bypass attempt.
- [PROMPT_INJECTION]: The skill processes untrusted task data, which represents an indirect prompt injection surface. Ingestion points: Task data from
lark-cli task +get-my-tasks(SKILL.md). Boundary markers: Not present in report generation templates. Capability inventory: Includes message sending and document creation vialark-cli. Sanitization: No validation or escaping of task content is performed.
Audit Metadata