Baby Blanket Kick Monitoring Skill | 婴儿蹬被监测技能

Identifies babies kicking off blankets or exposing their bodies during sleep and alerts parents to cover them up to prevent catching a cold. | 婴儿蹬被监测技能，识别婴儿夜...

Key points to consider before installing or using this skill: - Source and privacy: The skill will accept and upload infant monitoring videos to external API endpoints configured in the code (smyx_common ApiEnum BASE_URLs). There is no homepage or official source; verify the API host (config.yaml defaults reference lifeemergence.com) and privacy policy before sending sensitive video/PII. - Hidden/implicit credentials: Although the registry lists no required environment variables, the code expects API keys/open-id either in skills/smyx_common/scripts/config.yaml, workspace config, or via command-line. Do not provide production credentials or any sensitive tokens unless you trust the code and the remote API. - Local storage and retention: The skill saves uploaded files to an attachments folder and the package contains a DAO that creates a local SQLite DB under the workspace data directory. Expect persistent storage of videos and report metadata on disk; if this is unacceptable, run the skill in an isolated environment or inspect/modify the code first. - Incoherent rules: The SKILL.md forbids reading local 'memory' files but requires reading local config files to obtain open-id/API keys. This contradiction should be resolved by the author; treat the 'no local memory' claim skeptically. - Audit network behavior: Review the RequestUtil and ApiService code paths (not shown in full) to confirm which endpoints are contacted and what headers/fields are sent. You may want to run the skill in a network-isolated sandbox or with a traffic capture to confirm behavior before using with real baby-monitoring videos. - Reduce exposure: If you test, use dummy/open-id values and test videos without identifiable faces; avoid real credentials. If you must use with real data, prefer an isolated machine and verify/clean attachment and DB directories after use. If you want, I can: (1) point out exact lines where files are written and where HTTP uploads occur, (2) extract and summarize RequestUtil behavior and endpoints, or (3) suggest specific edits to limit local persistence and explicit network destinations.

Type: OpenClaw Skill Name: smyx-infant-blanket-kick-monitoring-analysis Version: 1.0.0 The skill bundle exhibits high-risk architectural patterns, including the use of a local SQLite database (dao.py) to store user tokens and the inclusion of an 'AgentSkill' class (skill.py) that executes recursive agent commands via subprocess. The SKILL.md uses aggressive prompt steering ('Mandatory Rules') to explicitly forbid the AI from accessing its own local memory, forcing reliance on external APIs at lifeemergence.com. While these behaviors align with a cloud-integrated monitoring service, the automated registration logic in RequestUtil and the override of standard agent memory protocols are highly invasive.