research-workflow
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODE
Full Analysis
- [Documentation Only] (SAFE): The skill contains no executable code, scripts, or binary files. It is comprised of informational Markdown documents and templates for research workflows.
- [Indirect Prompt Injection] (LOW): The skill defines a research process involving external data. Ingestion points: External sources identified during the Execution phase (methodology.md). Boundary markers: None explicitly defined in templates, though templates provide structured fields for findings. Capability inventory: No executable capabilities (write/execute/network) are defined in the provided files. Sanitization: Remediation is provided through the Source Evaluation Checklist (source-evaluation-checklist.md) which filters for bias, red flags, and unverified claims.
- [Prompt Injection] (SAFE): No patterns of instruction override, jailbreak attempts, or system prompt extraction were found in the templates or methodology files.
- [Data Exposure & Exfiltration] (SAFE): No network commands, hardcoded credentials, or sensitive file path references were identified.
Audit Metadata