Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Lovart API Skills
v1.0.0Generate images, videos, and audio/music via Lovart AI. Also manages Lovart projects, threads (conversation history), and user settings. Trigger on: (1) any...
⭐ 0· 22·0 current·0 all-time
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 env vars (LOVART_ACCESS_KEY/LOVART_SECRET_KEY), and python3 dependency align with a Lovart API client. However the registry metadata says 'instruction-only' but a sizable Python client (agent_skill.py) is included — that mismatch is unexpected. The SKILL.md also instructs the agent to use built-in commands only (no direct API calls) while the shipped Python client makes direct HTTP calls to Lovart endpoints (this may be intended, but the documentation forbids inspecting the code).
Instruction Scope
SKILL.md requires checking local state (~/.lovart/state.json) before first generation and mandates downloading and sending local files created under /tmp. The registry's 'required config paths' lists none, so the instructions ask the agent to access a specific user file that was not declared. The skill also forbids inspecting its own source code (agent_skill.py), which prevents verifying endpoints or behavior — that restriction plus undisclosed file access is suspicious.
Install Mechanism
There is no install spec (instruction-only) which is low friction, but a non-trivial Python client file is bundled. No external downloads or package installs are requested, which lowers supply-chain risk, yet including executable code without an install step means the code will run in-agent — review required.
Credentials
Only LOVART_ACCESS_KEY and LOVART_SECRET_KEY are required and the primaryEnv is one of them — this is proportionate for an API client that uses HMAC signing. No unrelated credentials are requested.
Persistence & Privilege
The skill does not request 'always: true' and is user-invocable, which is normal. It does read/write local files (reads ~/.lovart/state.json, writes downloads under /tmp/openclaw) and will send downloaded files as attachments — this is functional for a media-generation client but has privacy implications. The Python client also disables SSL certificate verification (lenient SSL context), which weakens network security.
What to consider before installing
This skill appears to implement a Lovart API client and asks only for the usual API keys, but there are red flags you should weigh: (1) SKILL.md requires reading ~/.lovart/state.json but the registry metadata doesn't declare any config path — expect the skill to read that home file; (2) a full Python client (agent_skill.py) is bundled even though the skill is described as instruction-only, and the document forbids inspecting that source — you cannot easily verify endpoints or behavior unless you open the file yourself; (3) the client disables SSL certificate verification, increasing risk of man-in-the-middle attacks when it talks to servers; (4) the skill will download artifacts to /tmp and must send those files to the user, so avoid giving it access to sensitive local files; (5) if you install, prefer using scoped or rotatable API keys, review the agent_skill.py code before granting keys, and avoid using high-privilege or long-lived credentials until you trust the source. If you cannot inspect or verify the code and the origin of this skill is unknown, treat it as higher-risk and consider not installing.Like a lobster shell, security has layers — review code before you run it.
latestvk978mhkezhfp2wp1j40mbtf3kx84kw3m
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🎨 Clawdis
Binspython3
EnvLOVART_ACCESS_KEY, LOVART_SECRET_KEY
Primary envLOVART_ACCESS_KEY