Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Closeli Open Device Event Query
v1.0.0调用 ai-open-gateway 的事件查询接口 POST /api/event/query，支持自然语言查询设备事件，返回 AI 摘要和事件列表。Use when: 需要查询设备检测到的事件、了解某段时间内的活动情况，例如有没有人出现、有没有车开过、我的猫去哪里了等自然语言问题。⚠️ 需设置 AI_GATE...
⭐ 0· 20·0 current·0 all-time
byCloseliOpenTeam@closeli-open
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description, required binary (python3), and the single required credential (AI_GATEWAY_API_KEY) are coherent with a device-event query skill. The script calls only the declared endpoint (/api/event/query) and uses Bearer auth as documented. Minor metadata mismatch: SKILL.md declares a fallback config path (~/.openclaw/.env) in its openclaw.requires.configPaths, but the top-level registry metadata says "Required config paths: none" — this is an inconsistency to be aware of.
Instruction Scope
SKILL.md enforces strict display/formatting rules (agent must reformat JSON, strip device_id prefix, produce Markdown links, limit displayed items) while the script purposely outputs raw JSON for the agent to parse. Functional mismatches: SKILL.md says "超过 10 条只展示前 10 条" (show up to 10) but the script truncates events to MAX_EVENTS=3. The device_id prefix removal and Markdown formatting are agent-side responsibilities (documented) but the instructions and implementation are not perfectly aligned.
Install Mechanism
No install spec (instruction-only) and a small Python script is provided. The script depends on the third-party httpx package; it will exit with a message if httpx is not installed (user must run pip). This is low to moderate risk but the dependency is not automatically installed and is not fetched from arbitrary URLs.
Credentials
Only AI_GATEWAY_API_KEY is required (appropriate), but the script will read a shared fallback file (~/.openclaw/.env) unless AI_GATEWAY_NO_ENV_FILE=true. That file is shared across skills and may expose credentials to other local skills; SKILL.md warns about this but the default behavior reads it. There is also an environment flag (AI_GATEWAY_VERIFY_SSL=false) that can disable TLS verification — dangerous if used in production. These behaviors increase risk if users do not follow the recommended mitigations.
Persistence & Privilege
Skill is not always-enabled and does not request elevated or persistent platform privileges. It does not modify other skills or system-wide settings. Autonomous invocation is allowed (platform default) but not combined with other high-risk factors here.
What to consider before installing
This skill appears to do what it says (call Closeli's ai-open-gateway /api/event/query) but take these precautions before installing: 1) Only provide a key with minimal (event-query) permissions. 2) Avoid putting your API key in ~/.openclaw/.env — prefer environment variables and set AI_GATEWAY_NO_ENV_FILE=true to disable the shared fallback. 3) Never set AI_GATEWAY_VERIFY_SSL=false in production. 4) Confirm the AI_GATEWAY_HOST points to a trusted domain. 5) Note the small mismatches: the script truncates events to 3 while SKILL.md describes a 10-item display rule, and SKILL.md requires agent-side formatting (prefix stripping/Markdown); ensure your agent will perform those steps rather than printing raw JSON. If you need absolute assurance, inspect or run the script in a controlled environment and test with a minimally-scoped API key before using it with production credentials.Like a lobster shell, security has layers — review code before you run it.
latestvk97cx9vasy308swdv8n3xc2pts84fj84
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
Binspython3
EnvAI_GATEWAY_API_KEY
Primary envAI_GATEWAY_API_KEY