autonomous-agent

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (medium risk: 0.30). The skill specifies an autonomous agent that automatically performs networked knowledge research, auto-executes and auto-heals in silent/never-stop mode, which meaningfully weakens user control and represents suspicious steering without explicit malicious intent.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). 技能的工作流匹配/知识调研层会在“无可用工作流”或需要调研时去“搜索 GitHub 开源项目/文章和教程并提取步骤”，从而会在运行时读取外部（第三方）来源的自由文本内容（如 GitHub Issues/README/教程页面段落）。