lead-researcher

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a procedural guide for complex research workflows. It contains no executable code, scripts, or automated network commands.
  • [EXTERNAL_DOWNLOADS]: The skill references architectural documentation from Anthropic's official website to guide the agent's behavior.
  • [INDIRECT_PROMPT_INJECTION]: The skill handles data from external research sources through sub-agents.
  • Ingestion points: Research findings gathered from the web and external sources by sub-agents (Phase 2).
  • Boundary markers: The lead agent is instructed to assign specific objectives and boundaries to sub-agents to limit scope.
  • Capability inventory: The skill involves writing synthesized reports to the local file system at ./research/.
  • Sanitization: The instructions mandate strict source attribution for every claim to ensure transparency and verifiability of the ingested data.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 01:00 PM
Security Audit — agent-trust-hub — lead-researcher