deep-research

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill consists purely of markdown instructions and organizational templates to guide an AI agent through a research workflow. There is no executable code or configuration that poses a security risk.
  • [NO_CODE]: No scripts, binaries, or automated shell commands are included in the skill content.
  • [PROMPT_INJECTION]: The skill workflow involves processing external research data in SKILL.md, representing a potential surface for indirect prompt injection. However, as it includes no code or automated tools to exploit this data, the risk is minimal. Ingestion points: External content gathered during research; Boundary markers: Absent; Capability inventory: None; Sanitization: The instructions recommend confidence marking and contradiction flagging as validation steps.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 03:00 PM