officecli-docx

Fail

Audited by Gen Agent Trust Hub on Oct 23, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides setup instructions that involve downloading and executing scripts directly in the shell (curl ... | bash and irm ... | iex) from the domain d.officecli.ai. While this is presented as a standard installation method for the vendor's officecli tool, it follows a remote code execution pattern.
  • [COMMAND_EXECUTION]: The skill relies on executing shell commands with various flags and arguments to manipulate Word documents. This involves passing user-defined or agent-generated content into command-line properties.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze data from external .docx files, which creates an attack surface for indirect prompt injection.
  • Ingestion points: Reads text, structures, and metadata from Word documents via commands such as officecli view text, officecli view html, and officecli get (SKILL.md).
  • Boundary markers: None; there are no specific instructions or markers used to distinguish between legitimate document data and potentially malicious embedded instructions.
  • Capability inventory: The skill uses the officecli tool which has broad capabilities to read, write, and modify files on the system based on the agent's interpretation of document content.
  • Sanitization: The instructions do not specify any sanitization, escaping, or validation of the content extracted from Word files before it is processed by the agent.
Recommendations
  • HIGH: Downloads and executes remote code from: https://d.officecli.ai/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Oct 23, 2026, 12:06 PM