Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

应用资产查询

v1.1.0

友盟应用资产查询 skill。当用户想知道自己在友盟有哪些应用、应用数量、小程序列表时使用。触发词:应用列表、我的应用、小程序列表、应用数量、有哪些应用、应用资产。

0· 57·0 current·0 all-time
byUmeng+@squall0925
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
Capability signals
Requires sensitive credentials
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
VirusTotalVirusTotal
Pending
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
Name/description: read-only app-asset queries against Umeng. Included files: a bundled Umeng OpenAPI Python SDK (many API classes, including create/edit endpoints) and scripts/assets.py as the entry. Including the SDK is reasonable for calling Umeng APIs, but the bundle contains write-capable API classes (create/edit) that are not required by the stated read-only purpose; this increases the attack surface though it may be benign for reuse.
!
Instruction Scope
SKILL.md instructs running python3 scripts/assets.py and describes configuration via --config or UMENG_CONFIG_PATH and local umeng-config.json. The manifest declares no required env vars/credentials, yet runtime clearly depends on a config file that likely contains appKey/secret or accessToken. The instructions do not explicitly declare where credentials come from or what the config contains. Because the runtime script will perform network calls to gateway.open.umeng.com, users need to know what data is read and transmitted; SKILL.md does not show that explicitly.
Install Mechanism
No install spec (instruction-only), so nothing is automatically downloaded at install time. However, the skill includes ~75 files (the SDK and scripts) that will be present on disk and executed when invoked. No external arbitrary-download URLs were used in the manifest.
!
Credentials
No required environment variables or primary credential declared in the registry metadata, but SKILL.md references UMENG_CONFIG_PATH and a umeng-config.json that will contain credentials (appKey/secret or access token). The skill omits declaring these required secrets, which is an incoherence: the agent/operator must provide credentials but the skill metadata doesn't advertise them. Storing secrets in a plaintext JSON file (as suggested) is also a potential security risk.
Persistence & Privilege
always is false and model invocation is allowed (platform default). The skill does not request permanent/global privileges or modify other skills' config. No other elevated persistence behavior is indicated.
What to consider before installing
This skill appears to be a legitimate Umeng app-listing helper, but there are important mismatches you should address before installing: - Credentials: The SKILL.md expects a umeng-config.json or UMENG_CONFIG_PATH containing your Umeng credentials, but the skill metadata does not declare required env vars or secrets. Do not provide global or highly privileged credentials without review. Inspect the sample umeng-config.json and scripts/assets.py to confirm exactly what keys/tokens are read and stored. - Review the code: Because the package bundles a full OpenAPI SDK (including create/edit API classes), review scripts/assets.py to ensure it only calls read/list/count APIs and does not perform modification or exfiltrate data to unexpected endpoints. - Secret storage: Avoid storing appKey/secret in plaintext in shared locations. Prefer using ephemeral credentials or a secrets manager and limit the permission scope of the credentials used. - Run in isolation: If you want to try it, run the script in a safe environment (isolated container or dedicated account) with only a read-only Umeng account or scoped credentials. If you want, I can: (1) show the contents of scripts/assets.py and the umeng-config.json example so you can verify what is read and sent; (2) list all API classes included to highlight any write-capable methods; or (3) suggest a minimal safe config layout and recommended permission scope for credentials.

Like a lobster shell, security has layers — review code before you run it.

latestvk9705dstj9sq6bdthc2zkbwn2h84xeac

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments