content-research-writer
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is composed entirely of instructional Markdown content, decision tables, and templates intended to guide an AI agent through a multi-phase research and drafting process. No shell commands, scripts, or tool calls are present in the provided file.
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a process for ingesting external data from academic databases and industry reports, creating a theoretical attack surface for indirect prompt injection from research materials. 1. Ingestion points: Discovery and extraction of findings from academic databases (Google Scholar, PubMed, etc.) and industry reports as described in Phase 2. 2. Boundary markers: None explicitly defined for technical separation of data, but the process mandates a thesis-driven outline (Phase 3) which acts as a logical constraint. 3. Capability inventory: The skill provides instructions for outlining, drafting, and citation management; it does not request or define file-writing, network, or subprocess tools within the SKILL.md file. 4. Sanitization: The skill includes a Source Evaluation Framework (CRAAP test) and a Fact-Checking Protocol that instructs the agent to verify all statistical claims and trace information back to primary sources before inclusion.
Audit Metadata