Trusted — Risk Score 5/100
Last scan:23 hr ago Rescan
5 /100
agent-link
智能体互联技能 - 支持不同电脑上的 OpenClaw 实例和 Agent 通过中转服务器进行安全可靠的通讯
This is a legitimate cross-device agent communication relay skill with no malicious behavior detected. All functionality is properly documented and uses standard WebSocket communication with HMAC-SHA256 message signing.
Skill Nameagent-link
Duration32.1s
Enginepi
Safe to install
This skill is safe to use. No security concerns identified.

Findings 1 items

Severity Finding Location
Low
Plain WebSocket connection default Sensitive Access
The skill defaults to ws:// (unencrypted) instead of wss:// (encrypted WebSocket). While WSS is recommended in documentation, the code defaults to non-encrypted connections which could expose message content in transit.
self.websocket = await websockets.connect(self.relay_url)
→ Consider defaulting to wss:// protocol or adding a warning when using unencrypted connections
scripts/local-agent/agent_link.py:54
ResourceDeclaredInferredStatusEvidence
Filesystem READ READ ✓ Aligned agent_link.py:188-195 - reads config JSON file via from_config()
Network READ WRITE ✓ Aligned agent_link.py:54-65 - WebSocket connection for sending/receiving messages
Shell NONE NONE No subprocess or shell execution found in codebase
Environment NONE NONE No os.environ access or sensitive variable reading found

File Tree

6 files · 25.3 KB · 1019 lines
Markdown 3f · 686L Python 1f · 315L JSON 2f · 18L
├─ 📁 docs
│ └─ 📝 install-agent.md Markdown 237L · 5.0 KB
├─ 📁 scripts
│ └─ 📁 local-agent
│ ├─ 🐍 agent_link.py Python 315L · 10.0 KB
│ └─ 📋 agent-link-config.example.json JSON 9L · 235 B
├─ 📋 _meta.json JSON 9L · 348 B
├─ 📝 README.md Markdown 219L · 4.9 KB
└─ 📝 SKILL.md Markdown 230L · 4.8 KB

Dependencies 1 items

PackageVersionSourceKnown VulnsNotes
websockets * pip No No version pinned - consider pinning to specific version for reproducibility

Security Positives

✓ HMAC-SHA256 message signing prevents message forgery
✓ No subprocess or shell execution - pure Python WebSocket client
✓ No credential harvesting or environment variable enumeration
✓ No sensitive file path access (~/.ssh, ~/.aws, .env)
✓ No base64 encoding or obfuscation detected
✓ All functionality is properly documented in SKILL.md
✓ Message relay server does not persist message content
✓ Clean, readable codebase with proper error handling