Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Skill
v1.1.0Security check for OpenClaw skills. Scan any ClawHub skill for malware, prompt injection, data theft, wallet stealing, and dangerous permissions BEFORE insta...
⭐ 0· 98·1 current·1 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The name/description match what the scripts do: query a remote trust database at checksafe.dev for a skill slug and present a badge/report. However, the SKILL.md claims this will 'automatically' pre-check before any clawhub install; there is no integration or hook provided to implement automatic pre-install checks (the repo only includes standalone shell scripts). The automatic pre-install behavior is therefore marketing rather than implemented functionality.
Instruction Scope
The runtime instructions and scripts call HTTPS endpoints at checksafe.dev with skill slugs and, in batch/scan mode, enumerate the local skills directory and send all discovered slugs to the remote API. The SKILL.md asserts 'Does not send any user data to Sentinel — only queries by skill slug', but scanning installed skills will in practice transmit the user's installed-skill inventory (slugs), which is user-specific information. There are no instructions that read or transmit local files or secrets, but the inventory leak is a privacy/leakage vector the docs understate.
Install Mechanism
No install spec and the shipped files are simple, readable shell scripts and markdown. Nothing is downloaded or executed from untrusted URLs by the scripts themselves (they only call checksafe.dev APIs). There is no archive download/extract behavior in the skill itself.
Credentials
The skill requests no environment variables or credentials, which aligns with its purpose. However, transmitting installed skill slugs to a third-party service is a form of data exfiltration (metadata) even if no secrets are sent; the SKILL.md's blanket statement that 'Does not send any user data to Sentinel' is inaccurate in the context of installed-skill scans.
Persistence & Privilege
The skill does not request always:true and does not modify other skills or system configuration. It can be invoked autonomously by an agent (default behavior) but that is normal for skills and not by itself a red flag here.
What to consider before installing
This skill appears to be a lightweight client for the external service checksafe.dev (it queries badge.json and report endpoints). Before installing, consider:
- Privacy: scan_installed.sh collects local skill slugs and sends them to checksafe.dev; if you are uncomfortable exposing your installed-skill inventory, do not run the batch/scan scripts or review them locally first.
- Trust the external service: the tool relies entirely on checksafe.dev. Verify the domain, TLS, and the service's privacy policy if you will send any inventory. If you need to avoid network calls, use the scripts only with explicit, manual slug names or inspect the API responses from a browser first.
- Automatic pre-install: the README/SKILL.md implies automatic pre-install hooks; those are not implemented. If you expect an agent to automatically intercept 'clawhub install', additional integration would be required.
- Confirm behavior: the code is readable and not obfuscated, so you can safely inspect scripts before running. If you want stronger guarantees, ask the author for an explanation of what metadata is logged server-side and for an option to run checks locally or provide an allowlist/offline dataset.
Given these mismatches (especially the inaccurate claim about not sending user data), treat the skill as useful but with privacy implications — review and run the scripts manually and confirm the external service's trustworthiness before enabling automatic or broad scans.Like a lobster shell, security has layers — review code before you run it.
latestvk976ng71k1y00kcermzamyjgyx83dnqb
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🛡️ Clawdis