Scan Report
5 /100
blender-mcp
Integrate Blender MCP (Model Context Protocol) allowing OpenClaw to control Blender for 3D modeling, scene creation, and manipulation
Blender MCP integration skill with properly declared capabilities - the execute_code feature is legitimate Blender automation and runs within Blender's sandboxed Python environment.
Safe to install
No action needed. The skill is safe to use.
Findings 1 items
| Severity | Finding | Location |
|---|---|---|
| Info | execute_code command declared Doc Mismatch | SKILL.md:31 |
| Resource | Declared | Inferred | Status | Evidence |
|---|---|---|---|---|
| Shell | WRITE | WRITE | ✓ Aligned | scripts/start-server.sh:9 - uvx blender-mcp |
| Network | READ | READ | ✓ Aligned | scripts/test-connection.py:17 - localhost:9876 only |
| Filesystem | NONE | NONE | — | No file operations in scripts |
File Tree
4 files · 7.2 KB · 226 lines Markdown 2f · 149L
Python 1f · 65L
Shell 1f · 12L
├─
▾
scripts
│ ├─
start-server.sh
Shell
│ └─
test-connection.py
Python
├─
README.md
Markdown
└─
SKILL.md
Markdown
Security Positives
✓ No credential harvesting - skill does not access ~/.ssh, ~/.aws, .env, or similar sensitive paths
✓ No network exfiltration - only communicates with localhost:9876 (Blender socket server)
✓ No base64 or obfuscated code present
✓ No reverse shell, C2, or data theft indicators
✓ All capabilities properly declared in SKILL.md
✓ Clean codebase with no suspicious patterns
✓ uvx is a standard Python tool runner with version-pinned dependencies