中文 EN

Scan Report

Scan New Files

⚠ Low Risk — Risk Score 28/100
Last scan:21 hr ago Rescan
28 /100
memory-orchestrator
全栈智能记忆系统 - AI agent semantic memory with emotion tagging, knowledge graphs, and self-evolution
A legitimate memory orchestration system with moderate security concerns: undeclared shell execution, supply chain risks from unpinned dependencies, and a documentation mismatch for the clipboard capability.
Skill Namememory-orchestrator
Duration68.5s
Enginepi
✓
Safe to install
Review and approve with conditions: pin all dependency versions, declare shell execution in documentation, and remove clipboard from capability map since it's not implemented.

Findings 5 items

Severity Finding Location
Medium
Unpinned dependency versions Supply Chain
requirements.txt uses version ranges (>=) instead of pinned versions (==), enabling dependency confusion and unexpected behavior
faiss-cpu>=1.7.4
sentence-transformers>=2.2.2
→ Pin all dependencies to specific versions to ensure reproducible builds and prevent supply chain attacks
 requirements.txt:3
Medium
Direct git repository installation Supply Chain
clip package is installed directly from GitHub URL, executing remote code during installation
clip @ git+https://github.com/openai/CLIP.git
→ Use a released version from PyPI or pin to a specific git commit hash
 requirements.txt:9
Low
System service manipulation via sudo Priv Escalation
install.sh uses sudo to enable and start syncthing@claw systemd service
sudo systemctl enable syncthing@claw
sudo systemctl start syncthing@claw
→ Ensure this is intentional for the skill's functionality and declare in documentation
 install.sh:51
Low
Cron job installation without explicit consent Persistence
install.sh silently adds a cron job to run auto-commit-memory.sh every 30 minutes
(crontab -l 2>/dev/null; echo "*/30 * * * * $SKILL_DIR/scripts/auto-commit-memory.sh") | crontab -
→ Make cron setup optional with --enable-cron flag or document clearly
 install.sh:68
Low
Capability declared but not implemented Doc Mismatch
SKILL.md capability map lists clipboard:READ but no code accesses clipboard functionality
Clipboard (clipboard) - Access: READ
→ Remove clipboard from capability map or implement if intended
 SKILL.md:1
ResourceDeclaredInferredStatusEvidence
Filesystem READ WRITE ✓ Aligned build_memory_index.py: Writes FAISS indexes and JSON metadata to $WORKSPACE/inde…
Network READ READ ✓ Aligned multimodal_processor.py: Downloads CLIP/Whisper models from HuggingFace
Shell NONE WRITE ✗ Violation install.sh:51 uses sudo systemctl; self_evolution_engine.py:200 subprocess.run f…
Environment NONE READ ✗ Violation Scripts read Path.home() and workspace paths from environment
Clipboard READ NONE ✗ Violation SKILL.md lists clipboard:READ but no code references clipboard access
Skill Invoke NONE READ ✓ Aligned Workflows reference other skills (memory-watchdog, agent-browser)
5 findings
🔗
Medium External URL 外部 URL
https://img.shields.io/badge/Clawhub-memory--orchestrator-blue
README.md:5
🔗
Medium External URL 外部 URL
https://clawhub.com/skills/memory-orchestrator
README.md:5
🔗
Medium External URL 外部 URL
https://img.shields.io/badge/version-1.0.0-green
README.md:6
🔗
Medium External URL 外部 URL
https://img.shields.io/badge/license-MIT-yellow
README.md:7
🔗
Medium External URL 外部 URL
https://ollama.ai
install.sh:31

File Tree

17 files · 147.6 KB · 4677 lines
Python 7f · 3014L YAML 4f · 1151L Markdown 2f · 282L Shell 3f · 206L Text 1f · 24L
├─ ▾ 📁 scripts
│ ├─ 🔧 auto-commit-memory.sh Shell 43L · 1.1 KB
│ ├─ 🐍 build_memory_index.py Python 127L · 3.9 KB
│ ├─ 🐍 conflict_detector.py Python 372L · 14.4 KB
│ ├─ 🐍 emotion_tagger.py Python 673L · 21.7 KB
│ ├─ 🐍 knowledge_graph_builder.py Python 341L · 12.4 KB
│ ├─ 🐍 multimodal_processor.py Python 730L · 22.5 KB
│ ├─ 🐍 search_memory.py Python 95L · 2.8 KB
│ └─ 🐍 self_evolution_engine.py Python 676L · 22.8 KB
├─ ▾ 📁 workflows
│ ├─ 📋 memory-activerecommend.yaml YAML 644L · 17.6 KB
│ ├─ 📋 memory-knowledgegraph.yaml YAML 137L · 4.5 KB
│ ├─ 📋 memory-multimodal.yaml YAML 230L · 6.8 KB
│ └─ 📋 memory-sync.yaml YAML 140L · 4.2 KB
├─ 🔧 install.sh Shell 107L · 3.5 KB
├─ 📝 README.md Markdown 88L · 2.0 KB
├─ 📄 requirements.txt Text 24L · 363 B
├─ 🔧 run.sh Shell 56L · 1.4 KB
└─ 📝 SKILL.md Markdown 194L · 5.6 KB

Dependencies 7 items

PackageVersionSourceKnown VulnsNotes
faiss-cpu >=1.7.4 pip No Version not pinned
sentence-transformers >=2.2.2 pip No Version not pinned
torch >=2.0.0 pip No Version not pinned
clip git+https://github.com/openai/CLIP.git git No Direct git installation - supply chain risk
whisper >=1.1.10 pip No Version not pinned
networkx >=3.1 pip No Version not pinned
transformers >=4.30.0 pip No Version not pinned

Security Positives

✓ No credential harvesting or API key theft detected
✓ No data exfiltration to external servers
✓ No base64-encoded or obfuscated code
✓ No reverse shell or C2 communication patterns
✓ subprocess calls only legitimate CLI tools (ollama, git, systemctl)
✓ Memory data stored locally in user's workspace
✓ git-crypt encryption for sensitive files is appropriate
✓ No hidden HTML comments or prompt injection attempts