officecli-academic-paper
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill provides instructions for interacting with a pre-installed CLI tool via shell commands. It enforces strict shell quoting rules to prevent command injection during the assembly of document properties from user-provided text.
- [EXTERNAL_DOWNLOADS]: The skill explicitly forbids the use of network tools like curl or external package managers, requiring the agent to use only pre-installed software within the environment.
- [INDIRECT_PROMPT_INJECTION]: The skill handles untrusted user input for document generation. \n
- Ingestion points: User-provided content for document fields including titles, author names, abstract text, and bibliography entries. \n
- Boundary markers: Explicit instructions to use single-quoting for properties to prevent shell interpolation. \n
- Capability inventory: Access to bash for running document creation and conversion tools (officecli, soffice). \n
- Sanitization: Requirement for manual quoting and verification of command arguments before execution.
- [DATA_EXFILTRATION]: All file handling is performed using platform-standard artifact management tools that restrict data flow between the user's workspace and the isolated execution sandbox.
Audit Metadata