researchwrite

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes a utility Python script (build_proposal_docx.py) intended to convert Markdown research drafts into properly formatted Word documents (.docx). This script operates locally and is a functional part of the research deliverable workflow.
  • [EXTERNAL_DOWNLOADS]: The skill documentation refers to standard dependencies including specific platform-based skills (brainstorming, professor, avoid-ai-writing) and the widely recognized python-docx library. These resources are consistent with the skill's purpose.
  • [INDIRECT_PROMPT_INJECTION]: As a writing aid, the skill ingests user-supplied text (research ideas, drafts). It mitigates potential instruction override risks through a highly structured multi-stage workflow, using specific Markdown templates and 'section contracts' that define allowed and forbidden claims, effectively segregating user data from agent instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 02:18 AM
Security Audit — agent-trust-hub — researchwrite