write-content
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection vulnerability surface due to its data processing model.
- Ingestion points: Web page data and SERP content from the top 5-10 competitor results are retrieved and analyzed in Phase 1 (Research) of
SKILL.mdand by the Researcher agent described inreferences/writing-pipeline.md. - Boundary markers: Absent; there are no explicit delimiters or strict isolation guidelines provided to instruct the model to disregard system commands embedded inside the retrieved competitor pages.
- Capability inventory: The skill is scoped to file operations for persisting business context (
business-context.md), reading references, and generating markdown content. No arbitrary command execution or subprocess creation tools are included. - Sanitization: Absent; external text content is evaluated directly for knowledge extraction and structural gap analysis without filtering or sanitizing text inputs.
Audit Metadata