phosphorons
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill generates an HTML specification template in references/gabarit-spec-fonctionnelle.md that references https://cdnjs.cloudflare.com/ajax/libs/mermaid/11.15.0/mermaid.min.js. This fetches the Mermaid.js library from Cloudflare's trusted CDN for client-side diagram rendering.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze external data, including project documentation, code structure, and web content, which introduces a potential surface for indirect prompt injection.
- Ingestion points: The skill reads READMEs, documentation, code arborescence, and external competitor links as specified in SKILL.md (Level 0) and EVALUATIONS.md (E-3).
- Boundary markers: The instructions do not specify explicit delimiters for untrusted content, but the workflow requires the agent to summarize all facts for user validation before proceeding.
- Capability inventory: The skill utilizes file reading (depot context), web searching (competitor research), file writing (HTML spec and Markdown journal), and the AskUserQuestion tool for interaction.
- Sanitization: In references/gabarit-spec-fonctionnelle.md, the agent is specifically instructed to escape special characters like < and & within diagram labels to prevent rendering errors or injection vulnerabilities.
Audit Metadata