japanese-corporate-pptx
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data, including user-provided files (PPTX, PDF, images) and web content retrieved during research phases. This creates an attack surface where malicious instructions embedded in these external sources could influence the agent's behavior.
- Ingestion points: SKILL.md specifies that the agent should inspect user-provided documents and conduct web research using search engines.
- Boundary markers: The instructions do not define specific delimiters or 'ignore' blocks to protect the agent prompt from instructions contained within the ingested reference materials.
- Capability inventory: The skill utilizes host-provided presentation authoring tools to read, edit, and export PPTX files, as noted in SKILL.md.
- Sanitization: There are no documented procedures for sanitizing or escaping content from retrieved documents or web pages before it is processed by the agent.
- [COMMAND_EXECUTION]: The qa-checklist.md file instructs the agent to perform validation of the generated PPTX package using the 'unzip -t' shell command. While this is a standard integrity check, it involves executing system-level commands on generated artifacts.
Audit Metadata