Scan Report
0 /100
turboquant-plus
TurboQuant+ KV cache compression for llama.cpp on Apple Silicon — 3.8-6.4x compression with minimal quality loss
Pure documentation skill for TurboQuant+ KV cache compression configuration with no executable code or suspicious behavior.
Safe to install
No action required. This skill is a documentation-only resource.
| Resource | Declared | Inferred | Status | Evidence |
|---|---|---|---|---|
| Filesystem | NONE | NONE | — | No file operations in skill |
| Network | NONE | NONE | — | External URLs are documentation references only |
| Shell | NONE | NONE | — | No shell scripts present |
| Environment | NONE | NONE | — | No environment access |
| Skill Invoke | NONE | NONE | — | No sub-skill invocations |
| Clipboard | NONE | NONE | — | No clipboard access |
| Browser | NONE | NONE | — | No browser operations |
| Database | NONE | NONE | — | No database access |
1 findings
Medium External URL 外部 URL
https://research.google/blog/turboquant-redefining-ai-efficiency-with-extreme-compression/ SKILL.md:10 File Tree
3 files · 6.8 KB · 159 lines Markdown 2f · 152L
JSON 1f · 7L
├─
package.json
JSON
├─
README.md
Markdown
└─
SKILL.md
Markdown
Security Positives
✓ No executable code present — pure documentation
✓ No dependencies or external packages
✓ No shell scripts or subprocess calls
✓ No credential or sensitive file access
✓ External URLs are legitimate references (Google Research, GitHub)
✓ No obfuscation or encoded payloads
✓ Clean package.json with empty dependencies