中文 EN

Scan Report

Scan New Files

✓ Trusted — Risk Score 5/100
Last scan:2 days ago Rescan
5 /100
openclaw-3d-blender-mcp
Instalación completa de Blender MCP para OpenClaw. Incluye setup local/remoto, ngrok, verificación, troubleshooting.
Blender MCP integration skill with clear documentation, legitimate subprocess usage for CLI tools, and standard TCP networking for remote Blender control. No malicious indicators detected.
Skill Nameopenclaw-3d-blender-mcp
Duration40.5s
Enginepi
✓
Safe to install
This skill is safe to use. All subprocess and network operations are declared in SKILL.md and are necessary for Blender MCP functionality.

Findings 1 items

Severity Finding Location
Low
Hardcoded ngrok endpoints in test scripts
Some test scripts (test_direct.py, mcp_client.py) contain hardcoded ngrok endpoints (8.tcp.ngrok.io:16325). While convenient for testing, these are not documented in SKILL.md.
HOST = '8.tcp.ngrok.io'
→ Consider using environment variables for host configuration, or document hardcoded endpoints in SKILL.md
 scripts/test_direct.py:12
ResourceDeclaredInferredStatusEvidence
Filesystem NONE NONE — No file writes detected
Network READ READ ✓ Aligned TCP connections to Blender via ngrok tunnel
Shell WRITE WRITE ✓ Aligned subprocess.Popen for uvx blender-mcp (documented in SKILL.md)
Environment NONE READ ✓ Aligned Only reads BLENDER_HOST, BLENDER_PORT, DISABLE_TELEMETRY
Skill Invoke NONE NONE — N/A
Clipboard NONE NONE — N/A
Browser NONE NONE — N/A
Database NONE NONE — N/A
1 findings
🔗
Medium External URL 外部 URL
https://www.youtube.com/watch?v=dxlyCPGCvy8
SKILL.md:20

File Tree

18 files · 65.6 KB · 2448 lines
Python 13f · 1455L Markdown 3f · 975L Shell 1f · 12L JSON 1f · 6L
├─ ▾ 📁 references
│ ├─ 📝 common_errors.md Markdown 492L · 11.4 KB
│ └─ 📝 coordinate_system.md Markdown 305L · 7.5 KB
├─ ▾ 📁 scripts
│ ├─ 🐍 blender_direct_v2.py Python 96L · 3.1 KB
│ ├─ 🐍 blender_direct.py Python 95L · 2.9 KB
│ ├─ 🐍 blender_wait.py Python 109L · 3.5 KB
│ ├─ 🐍 get_scene_info.py Python 104L · 2.8 KB
│ ├─ 🐍 http_bridge.py Python 111L · 3.2 KB
│ ├─ 🐍 mcp_client_full.py Python 152L · 4.5 KB
│ ├─ 🐍 mcp_client.py Python 77L · 2.1 KB
│ ├─ 🐍 mcp_persistent.py Python 114L · 2.9 KB
│ ├─ 🔧 start-server.sh Shell 12L · 258 B
│ ├─ 🐍 test_direct.py Python 88L · 2.5 KB
│ ├─ 🐍 test_modal.py Python 114L · 2.7 KB
│ ├─ 🐍 test_simple.py Python 66L · 2.3 KB
│ ├─ 🐍 test-connection.py Python 65L · 2.2 KB
│ └─ 🐍 verify_blender_connection.py Python 264L · 7.5 KB
├─ 📋 _meta.json JSON 6L · 296 B
└─ 📝 SKILL.md Markdown 178L · 4.0 KB

Dependencies 1 items

PackageVersionSourceKnown VulnsNotes
uvx any system No CLI tool, not a Python package

Security Positives

✓ All subprocess usage declared in SKILL.md via 'uvx blender-mcp' commands
✓ No credential harvesting or sensitive path access (.ssh, .aws, .env)
✓ No base64 encoding, eval(), or obfuscated code patterns
✓ No curl|bash or wget|sh remote script execution patterns
✓ No hidden instructions in HTML comments or elsewhere
✓ TCP networking is standard for Blender MCP/ngrok tunneling
✓ Environment variables used are relevant to Blender MCP (BLENDER_HOST, BLENDER_PORT, DISABLE_TELEMETRY)
✓ No data exfiltration or C2 communication patterns
✓ Legitimate CLI tool usage (subprocess for uvx blender-mcp)
✓ Well-structured, documented skill with clear purpose