Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
baiyin-cover-train-skill
v1.0.2当用户希望通过百音开放平台训练 AI 歌手模型、查询训练任务状态，或根据已有 taskId 返回最终训练结果时使用。
⭐ 0· 59·0 current·0 all-time
bybaiyin@jiuping520
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
Capability signals
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
OpenClaw
Suspicious
high confidencePurpose & Capability
The SKILL.md requires a BAIYIN_OPEN_KEY and describes calling Baiyin endpoints (ai.hikoon.com) which fits the stated purpose. However, the skill manifest declares no required env vars or primary credential — that mismatch indicates the manifest under-declares sensitive requirements. The SKILL.md also demands reading _meta.json and contacting a remote 'SkillHub' to check/update the skill, behavior not implied by the skill description.
Instruction Scope
Instructions mandate reading the local _meta.json and performing a mandatory remote version check before any business logic; if a newer version exists the agent must 'update local skill' before proceeding. The SkillHub endpoint and update mechanism are unspecified. This gives the remote service control over executing updated instructions and may result in arbitrary code/behavior being introduced at runtime. Other instructions (uploading files to the Baiyin file API and calling training/task endpoints) are coherent with purpose.
Install Mechanism
No install spec is provided (instruction-only), but the mandatory version-check + 'update local skill' implies the agent should download and write new skill files or trigger an install process. Because no update URL, signing, or verification is specified, this is a supply-chain risk: arbitrary code could be written/executed without clear provenance.
Credentials
SKILL.md explicitly requires BAIYIN_OPEN_KEY and uses Authorization Bearer for uploads and API calls, but the registry lists no required env vars or primary credential. The missing declaration is a red flag: the skill needs an API key correlated to its purpose but the manifest fails to record it. There are no other unrelated credentials requested.
Persistence & Privilege
always is false (good), but the required pre-check that updates the local skill grants the skill effective permission to alter its own files at runtime. That allows persistent changes to behavior and broadens the attack surface, especially since update sources and integrity checks are unspecified and failures are to be silenced.
What to consider before installing
Key issues: (1) The skill text requires a BAIYIN_OPEN_KEY for API calls but the manifest does not declare this — expect the agent to ask you for an API key before acting. (2) Before doing anything the skill mandates a remote 'SkillHub' version check and, if newer, to update the local skill; the SKILL.md does not specify where to fetch updates from, how updates are authenticated, or what 'update' entails. That is a supply-chain and persistence risk because it can change the skill's behavior silently. (3) The skill will upload user audio/images to https://ai.hikoon.com using your API key — consider privacy of those assets.
Recommendations before installing: verify the publisher identity and source (there is no homepage), insist the manifest explicitly declare required env vars (BAIYIN_OPEN_KEY) and the SkillHub/update endpoint, and confirm how updates are signed/authorized. Prefer running this skill in a restricted/sandboxed environment and do not provide sensitive credentials until you validate the service. If you must use it, require the agent to prompt before performing any remote update and to surface version-check network calls and their endpoints for review. Additional information that would change the assessment: a transparent, authenticated update mechanism (signed releases, explicit update URL), manifest updated to list BAIYIN_OPEN_KEY as required, and a public homepage/source repository.Like a lobster shell, security has layers — review code before you run it.
latestvk97fcmkf7bxmtc5gq9q6p9hhsh84ynm0
License
MIT-0
Free to use, modify, and redistribute. No attribution required.