baoyu-cover-image
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands such as
test -fto verify the presence of configuration files (EXTEND.md) and uses file system operations to manage reference images and prompt files. While these are common for local task management, they involve direct interaction with the host's shell and file system. - [PROMPT_INJECTION]: The skill processes untrusted article content and user-supplied metadata which is then embedded into prompts for downstream image generation tools. This represents an indirect prompt injection surface.
- Ingestion points: External article files, pasted text, and user-provided titles/summaries (e.g.,
Step 1: Analyze Content). - Boundary markers: The skill structures the generation prompt using Markdown, but does not specify robust sanitization or delimiters to prevent potentially malicious instructions within the article content from influencing the agent's behavior.
- Capability inventory: The skill allows the agent to write files, perform shell-based file checks, and invoke other image generation tools.
- Sanitization: No explicit sanitization or filtering of the article content is prescribed; instructions actually encourage using source text exactly as provided.
Audit Metadata