improve-article
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and transform untrusted user content (drafts, bullets, notes) into revised prose. This creates a surface where a malicious user could embed instructions to bypass constraints or manipulate the output.
- Ingestion points: The skill accepts user-provided articles, drafts, and notes in SKILL.md (Phase 2).
- Boundary markers: The skill uses delimiters like
--- Section [N] of [Total] — Original ---to separate user content during display, but lacks explicit instructions for the model to ignore embedded commands within that content. - Capability inventory: The skill performs natural language manipulation only; no file-writing, network communication, or shell command tools are used.
- Sanitization: No sanitization of input text is mentioned in the instructions.
Audit Metadata