lark-workflow-doc-perm

Pass

Audited by Gen Agent Trust Hub on Apr 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the lark-cli command-line tool to perform authenticated operations within the Lark/Feishu ecosystem. These operations include wiki node resolution (wiki spaces get_node), user directory searches (contact +search-user), and document permission modification (drive permission.members create) as described in SKILL.md.
  • [PROMPT_INJECTION]: The instructions incorporate emphatic formatting ('CRITICAL', 'MUST') to ensure the agent follows specific setup requirements for authentication. Additionally, the skill includes an indirect prompt injection surface by processing user-provided document tokens and names.
  • Ingestion points: User-supplied document identifiers, wiki URLs, and contact names processed in SKILL.md.
  • Boundary markers: No explicit delimiters or sanitization warnings are present to separate user data from operational instructions.
  • Capability inventory: The skill possesses the capability to modify document permissions and send instant messages via lark-cli (SKILL.md).
  • Sanitization: The skill does not implement specific validation or escaping for external content before passing it to tool parameters.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 13, 2026, 09:52 AM