中文 EN

Scan Report

Scan New Files

✓ Trusted — Risk Score 5/100
Last scan:1 day ago Rescan
5 /100
pharma-ai
智能药物发现AI助手，提供分子毒性预测、ADMET评估和虚拟筛选功能
PharmaAI is a legitimate drug discovery assistant using RDKit/scikit-learn for molecular toxicity prediction with no malicious behavior detected. Shell execution is documented and necessary for ML integration.
Skill Namepharma-ai
Duration44.4s
Enginepi
✓
Safe to install
Skill is safe for use. Consider pinning Python dependency versions for better supply chain hygiene.

Findings 2 items

Severity Finding Location
Low
Unpinned Python dependencies Supply Chain
requirements.txt uses >= version specifiers without upper bounds, allowing potentially incompatible versions to be installed
rdkit>=2023.0.0
→ Pin exact versions: rdkit==2023.9.0
 python-core/requirements.txt:1
Info
Missing screen.py implementation Doc Mismatch
python-bridge/index.ts calls virtualScreen which references a 'screen' script that does not exist in python-core/
await callPython('screen', {...})
→ Either implement screen.py or remove the virtualScreen export
 src/python-bridge/index.ts:80
ResourceDeclaredInferredStatusEvidence
Filesystem READ READ ✓ Aligned predict.py loads models from MODELS_DIR
Network NONE NONE — No network calls in codebase
Shell WRITE WRITE ✓ Aligned python-bridge/index.ts:17 spawns python3 subprocess
Environment NONE NONE — No os.environ access for sensitive data
Skill Invoke NONE NONE — Standard skill interface
Clipboard NONE NONE — Not used
Browser NONE NONE — Not used
Database NONE NONE — Not used
6 findings
🔗
Medium External URL 外部 URL
https://clawhub.com
HEADLESS_LOGIN.md:13
🔗
Medium External URL 外部 URL
https://clawhub.com/settings/tokens
HEADLESS_LOGIN.md:59
🔗
Medium External URL 外部 URL
https://docs.clawhub.com
HEADLESS_LOGIN.md:122
🔗
Medium External URL 外部 URL
https://discord.gg/clawd
HEADLESS_LOGIN.md:123
🔗
Medium External URL 外部 URL
https://docs.openclaw.ai/skills
PUBLISH_GUIDE.md:134
📧
Info Email 邮箱地址
[email protected]
HEADLESS_LOGIN.md:121

File Tree

12 files · 22.5 KB · 957 lines
Markdown 4f · 444L TypeScript 4f · 294L Python 1f · 163L JSON 2f · 52L Text 1f · 4L
├─ ▾ 📁 python-core
│ ├─ 🐍 predict.py Python 163L · 4.8 KB
│ └─ 📄 requirements.txt Text 4L · 64 B
├─ ▾ 📁 src
│ ├─ ▾ 📁 commands
│ │ └─ 📜 predict.ts TypeScript 77L · 2.2 KB
│ ├─ ▾ 📁 python-bridge
│ │ └─ 📜 index.ts TypeScript 112L · 2.4 KB
│ ├─ ▾ 📁 types
│ │ └─ 📜 index.ts TypeScript 47L · 920 B
│ └─ 📜 index.ts TypeScript 58L · 1.3 KB
├─ 📝 HEADLESS_LOGIN.md Markdown 126L · 2.8 KB
├─ 📋 package.json JSON 34L · 762 B
├─ 📝 PUBLISH_GUIDE.md Markdown 139L · 2.9 KB
├─ 📝 README.md Markdown 70L · 1.4 KB
├─ 📝 SKILL.md Markdown 109L · 2.5 KB
└─ 📋 tsconfig.json JSON 18L · 460 B

Dependencies 4 items

PackageVersionSourceKnown VulnsNotes
rdkit >=2023.0.0 pip No Version not pinned
scikit-learn >=1.3.0 pip No Version not pinned
numpy >=1.24.0 pip No Version not pinned
joblib >=1.3.0 pip No Version not pinned

Security Positives

✓ No network egress or C2 communication detected
✓ No credential harvesting or sensitive data access
✓ No obfuscated code or base64 execution
✓ Shell execution (python3 subprocess) is documented and necessary for ML integration
✓ No curl|bash or remote script execution
✓ File access limited to local model directory
✓ Clean, well-structured pharmaceutical ML codebase