Skills
Navigate
skills/dream-num/workbuddy-univer-office/univer-office/Gen Agent Trust Hub

univer-office

Pass

Audited by Gen Agent Trust Hub on Oct 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from various office file formats (XLSX, DOCX, PPTX) through tools like univer_import. This creates a vulnerability surface where malicious instructions hidden within document content could attempt to manipulate the agent's actions during inspection or comparison tasks.\n
  • Ingestion points: Document content is processed via univer_import and specific unit read operations in references/.\n
  • Boundary markers: The instructions do not mandate the use of delimiters or ignore-behavior prompts for external data.\n
  • Capability inventory: The agent can execute code via univer_execute and write to the filesystem via univer_export.\n
  • Sanitization: No sanitization or validation of the document content is specified before processing.\n- [DYNAMIC_EXECUTION]: The skill uses the univer_execute tool to run agent-generated code against the Univer Facade API. This represents a dynamic execution mechanism used to programmatically modify office documents at runtime.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 17, 2026, 02:11 PM