Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
LrshuAI First Tail Frame To Video
v1.0.2首尾帧生视频技能。当你需要提供首帧图片和尾帧图片来控制视频的起止画面并生成视频时调用。
⭐ 0· 42·0 current·0 all-time
by@lrshu
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name and description (first/tail-frame → video) align with the included script: the script uploads images/videos and prompts to a remote model API and polls for results. Requested python and TEAM_API_KEY are reasonable for calling a hosted API. However, the script defaults to a non-obvious backend (https://dlazy.com/api/ai/tool) and the SKILL.md does not document that backend or provide a trustable homepage/source, which merits verification.
Instruction Scope
SKILL.md's runtime instructions force the agent to execute `python script/invoke_model.py` directly and explicitly forbid using `openclaw run`. That bypass direction is unusual and potentially bypasses platform-level sandboxing/telemetry/policies. The script reads local files (images/videos), base64-encodes them and sends them to a remote endpoint using the TEAM_API_KEY bearer token — behavior consistent with the skill's purpose but also capable of exfiltrating arbitrary local files if misused. The SKILL.md/systemPrompt gives the agent strong, prescriptive runtime control which is a risk.
Install Mechanism
There is no install spec (instruction-only + included Python script), so nothing is pulled from external installers at install time. This is lower install risk. The script depends on the Python 'requests' module but no dependency declaration is provided; that may cause runtime errors but not an inherent security elevation.
Credentials
Only TEAM_API_KEY is required and used as a Bearer token to call the API, which is proportionate for a hosted-model integration. The script also honors an optional TEAM_BASE_URL env var (not documented in SKILL.md metadata) — the default backend (dlazy.com) is unfamiliar and should be validated. Confirm what privileges the TEAM_API_KEY grants (can it access other data/services?) before providing it.
Persistence & Privilege
The skill is not set to always:true and is user-invocable (normal). It does not request system-wide config changes or persistent elevated privileges. The main privilege concern is behavioral: the skill instructs direct execution of a script outside the platform runner, which increases its effective capability at runtime but is not expressed as a persistence flag.
What to consider before installing
This skill generally does what it claims (upload first/last frames to a remote model) but has several red flags you should address before installing: 1) Verify the backend: the script defaults to https://dlazy.com/api/ai/tool — confirm you trust that domain and ask the author for a homepage/source. 2) Review the code yourself or have it audited: the script will read local files and send base64-encoded contents to the remote API using your TEAM_API_KEY. 3) Do NOT supply a high-privilege TEAM_API_KEY until you confirm what that key can access. 4) Ask why the SKILL.md forbids `openclaw run` and forces direct `python` execution; prefer running the script inside a sandboxed environment or after the platform verifies its safety. 5) If you still want to use it, consider running it in an isolated environment, and set TEAM_BASE_URL explicitly to a vetted endpoint (do not rely on the default).Like a lobster shell, security has layers — review code before you run it.
latestvk977scwzgb8cdft3wrf5zm8zmn844phw
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🎞️ Clawdis
Binspython
EnvTEAM_API_KEY
Primary envTEAM_API_KEY