genshijin-crew

Pass

Audited by Gen Agent Trust Hub on May 11, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: The skill is a set of instructions for managing agent-to-subagent coordination and does not ship with any executable code, scripts, or external dependencies. It follows established vendor patterns for token optimization.- [PROMPT_INJECTION]: The skill outlines a workflow where a main agent interprets results from subagents, which represents an indirect prompt injection surface. However, the skill includes explicit instructions for maintaining safety by switching to standard language for security-sensitive warnings or ambiguous content.
  • Ingestion points: The agent processes 'tool-result' output from compressed subagents (SKILL.md).
  • Boundary markers: Structured format templates such as '' and 'path:line-range' are defined to delineate data (SKILL.md).
  • Capability inventory: The 'genshijin-builder' subagent is described as having file-editing capabilities (SKILL.md).
  • Sanitization: The '自動解除 (継承)' section requires agents to switch to standard language for security warnings or verification-heavy tasks (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
May 11, 2026, 06:00 AM