Skill flagged — suspicious patterns detected

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

lobster-rank

v2.0.1

Scan locally installed OpenClaw skills, collect evidence data, submit to the lobster ranking server for scoring, and let the user confirm uploading their res...

0· 52·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
high confidence
Purpose & Capability
The skill name/description (collect skill metadata and submit to a leaderboard) matches the code and runtime instructions: it scans skill directories, extracts structural metadata (names, flags, file counts, brief description lines), detects heuristics, and POSTs a JSON payload to the stated server. However, registry metadata supplied with the package incorrectly lists no required env vars/config paths while SKILL.md and the script do require/attempt to resolve an OPENCLAW_API_KEY (env or config). This metadata mismatch is an incoherence to be aware of.
Instruction Scope
The SKILL.md and the script stay within the declared purpose: scanning configured OpenClaw skill directories, reading SKILL.md files for simple description lines and keywords, checking for the presence (but not contents) of log files, reading openclaw.json only to extract an apiKey, and building a metadata payload. The instructions do not attempt to read arbitrary file contents, user secrets (beyond the apiKey), or transmit full file contents.
Install Mechanism
There is no install spec (no packages downloaded); the repository includes a client script (scripts/lobster_submit.py) which is intended to be run by the user. No high-risk external installers or archive downloads are used.
!
Credentials
The script requires an API key (OPENCLAW_API_KEY or CLI arg or value from openclaw.json). The package registry metadata incorrectly declared no required env/config, creating a transparency gap. More importantly, the default server URL is an HTTP URL to a numeric IP (http://47.94.254.45/api), meaning API keys and payloads will be transmitted unencrypted over the network unless the user overrides the URL. Using a raw IP and plaintext HTTP raises confidentiality and authenticity concerns for the API key and uploaded metadata.
Persistence & Privilege
The skill does not request 'always: true', does not modify other skills, and does not require elevated system privileges. It reads user-local skill directories and config paths that are reasonable for the stated purpose. Autonomous invocation is enabled by default (normal for skills) and is not combined with other red flags to escalate concern.
What to consider before installing
This skill appears to be what it claims—collecting structural metadata about your installed OpenClaw skills and submitting that metadata to a leaderboard server—but proceed with caution: (1) The package metadata advertised no required env/config but the script does need an OPENCLAW_API_KEY (it can also read apiKey from your openclaw.json). That mismatch is a transparency issue. (2) The default API endpoint is a numeric IP over plain HTTP (http://47.94.254.45); submitting the API key and metadata will be unencrypted and unauthenticated unless you override the URL to a secure endpoint. (3) The server is operated by the skill publisher (per SKILL.md), so you will be sharing skill names, counts, and small description/heuristic flags with that operator. Recommended actions: 1) Inspect scripts/lobster_submit.py yourself (you have it) before running. 2) Run in dry-run mode first (--dry-run) to see exactly what would be collected. 3) Do not paste your real OPENCLAW_API_KEY unless you trust the operator; prefer a temporary or read-only key if available. 4) Ask the publisher or maintainer to provide an HTTPS URL and a domain name (not a naked IP) and to fix the registry metadata to declare the required env/config. If you are uncomfortable with these issues, do not confirm uploads to the leaderboard.

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

latestvk97b6x0x7tvpm6m4atvv5jqsxx84abz9

License

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

Comments