Claude Code 控制器

• Analyzed: Mar 26, 2026, 8:47 AM
• Findings: 1
• Version: 1.0.0
• Hash: 05fb904f8c30169d1ee7fc940abb9d21f5509266289b0fc9706d5f62bf0384d4
• Source: palm
• External report: View on VirusTotal

Findings (1)

The skill provides instructions for an AI agent to control the Claude Code CLI, specifically recommending the use of the `--permission-mode bypassPermissions` flag to automate tasks without manual approval. While the skill includes safety guidelines (e.g., avoiding home and configuration directories) and installs the legitimate `@anthropic-ai/claude-code` package, the explicit instruction to bypass security prompts in a tool capable of arbitrary shell execution constitutes a high-risk capability. No evidence of intentional malice, data exfiltration, or obfuscation was found in SKILL.md or _meta.json.