marp-slide-creator
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [Unverifiable Dependencies & Remote Code Execution] (LOW): The technical slide template in
references/template.mdincludes a<script>tag that fetches the Mermaid library fromhttps://cdn.jsdelivr.net/npm/mermaid@latest/dist/mermaid.esm.min.mjs. Use of the@latestversion tag is unverifiable and could lead to issues if the upstream library is compromised. The severity is set to LOW as this is a standard and intended mechanism for enabling diagrams in Marp slides. - [Indirect Prompt Injection] (LOW): The skill ingests user requirements and outlines to generate presentation content, which creates a surface for indirect instructions to influence agent behavior.
- Ingestion points: User-provided purpose, outline, and slide content collected as described in
SKILL.md. - Boundary markers: Absent; there are no specific delimiters or instructions for the agent to treat input exclusively as data.
- Capability inventory: Capability to write files (
slides.md) and include potentially executable HTML/JavaScript tags in the output. - Sanitization: Absent; no explicit sanitization or validation of user-provided strings is mentioned before interpolation into the Markdown output.
Audit Metadata