multi-platform-publisher
Pass
Audited by Gen Agent Trust Hub on Jul 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a bundled Python script (
prepare_article.py) to parse article metadata and content. It also instructs the agent to usepandocfor converting Markdown files to DOCX or HTML formats to ensure compatibility with various platform editors. These are legitimate operations for the skill's stated purpose. - [EXTERNAL_DOWNLOADS]: As part of the publishing workflow, the agent may download remote image assets referenced in the article's Markdown to local temporary storage before uploading them to platforms that require direct file uploads. These downloads are directed by the content of the article provided by the user.
- [PROMPT_INJECTION]: The skill processes untrusted user-provided Markdown files and interpolates the extracted content into platform payloads and browser inputs. While it performs structure validation (e.g., checking for unclosed code fences), it lacks explicit boundary markers for the article body, making it susceptible to indirect prompt injection. This is a low-level risk inherent to the task of content processing.
Audit Metadata