essay-publishing-pipeline
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external markdown files which may contain embedded instructions designed to hijack the agent's behavior during the conversion or publishing process.
- Ingestion points: Markdown files are read from a source directory using
Path(source_dir).glob("*.md")andmd_file.read_text()inSKILL.md. - Boundary markers: The skill lacks explicit instructions or boundary markers to differentiate between the markdown content being processed and the instructions for the agent.
- Capability inventory: The skill uses
markdown.markdown()for HTML conversion, writes files to the filesystem usingoutput.write_text(), and interacts with external APIs (Dev.to and Newsletter) in thecross_postfunction. - Sanitization: There is no evidence of sanitization for the markdown content or frontmatter before it is interpolated into templates or sent to external APIs.
Audit Metadata