lark-attendance
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill involves the execution of the
lark-clitool to interact with the Lark API. The agent is instructed to build and run commands based on schema results and user-provided inputs.\n- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes data from external tool outputs without explicit sanitization or boundary markers.\n - Ingestion points: Data returned by
lark-cli schemaand attendance query results inSKILL.md.\n - Boundary markers: Absent; the skill does not provide instructions to delimit or ignore potential commands within tool output.\n
- Capability inventory: The skill enables command execution through
lark-cli.\n - Sanitization: Absent; the skill does not specify any methods for filtering or validating the output from the CLI tool.
Audit Metadata