中文 EN

Scan Report

Scan New Files

⚠ Low Risk — Risk Score 25/100
Last scan:1 day ago Rescan
25 /100
ratemyclaw
Score your OpenClaw agent setup against similar agents
This is a legitimate workspace-scoring tool with transparent data handling. The main concern is unpinned dependencies and unverified model downloads, but no malicious behavior or credential theft was found.
Skill Nameratemyclaw
Duration37.1s
Enginepi
✓
Safe to install
Pin sentence-transformers to a specific version and consider adding model hash verification for production use.

Findings 3 items

Severity Finding Location
Medium
Unpinned sentence-transformers dependency Supply Chain
SKILL.md instructs `pip install sentence-transformers` without version pinning. This allows a malicious version to be installed in the future.
pip install sentence-transformers
→ Pin to a specific version: pip install sentence-transformers==2.7.0
 SKILL.md:42
Low
Model download without integrity verification Supply Chain
The script downloads the all-MiniLM-L6-v2 model (~80MB) from HuggingFace Hub without hash verification. If the model server is compromised, a malicious model could be executed locally.
model = SentenceTransformer(EMBEDDING_MODEL)
→ Add model hash verification: SentenceTransformer(EMBEDDING_MODEL, model_kwargs={'cache_dir': cache, 'use_safetensors': True})
 scripts/submit_profile.py:95
Low
Scans .secrets directory for file names Sensitive Access
The script scans the .secrets directory and records file names. While it does NOT read secret contents (only names), this could reveal existence of credentials to an external service.
elif subdir == '.secrets': signals['secrets'] = [f.name for f in items]
→ This is documented behavior and reasonable for the use case. No action required.
 scripts/profile_generator.py:227
ResourceDeclaredInferredStatusEvidence
Filesystem READ READ ✓ Aligned profile_generator.py scans workspace with read-only operations
Network WRITE WRITE ✓ Aligned submit_profile.py makes POST requests to ratemyclaw.com
Shell NONE NONE — No subprocess or shell execution found
Environment READ READ ✓ Aligned Reads RATEMYCLAW_API_KEY from os.environ
Skill Invoke NONE NONE — No skill invocation detected
Clipboard NONE NONE — No clipboard access found
3 findings
🔗
Medium External URL 外部 URL
https://ratemyclaw.com**
README.md:7
🔗
Medium External URL 外部 URL
https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2
README.md:22
🔗
Medium External URL 外部 URL
https://ratemyclaw.com
SKILL.md:7

File Tree

5 files · 39.5 KB · 1147 lines
Python 2f · 732L JSON 1f · 259L Markdown 2f · 156L
├─ ▾ 📁 references
│ └─ 📋 taxonomy.json JSON 259L · 4.6 KB
├─ ▾ 📁 scripts
│ ├─ 🐍 profile_generator.py Python 539L · 22.6 KB
│ └─ 🐍 submit_profile.py Python 193L · 6.3 KB
├─ 📝 README.md Markdown 63L · 2.2 KB
└─ 📝 SKILL.md Markdown 93L · 3.8 KB

Dependencies 1 items

PackageVersionSourceKnown VulnsNotes
sentence-transformers * pip No Version not pinned - supply chain risk

Security Positives

✓ Embedding generation happens locally - raw text never leaves the machine
✓ API key is protected with chmod 600 after creation
✓ No raw file contents are transmitted, only taxonomy tags
✓ User confirmation is required before generating API key
✓ No shell execution (subprocess) detected
✓ No credential theft beyond the tool's own API key
✓ No base64 encoding, obfuscation, or suspicious patterns
✓ Good documentation with clear data flow explanation
✓ Secrets directory scanning only records names, not contents