Pyx Scan

Check whether an AI agent skill is safe before installing or using it. Calls the PYX Scanner API to retrieve trust status, risk score, and safety recommendat...

This skill simply asks the PYX Scanner service (https://scanner.pyxmate.com) about a skill identified as owner/name and formats the returned report. Before installing or relying on this skill, consider: 1) Do you trust the external PYX Scanner service and its privacy practices? The skill will make outbound requests to that domain for every check. 2) Avoid using it to scan private/internal skills if you don't want to share their owner/name with an external service. 3) The runtime has permission to run curl/WebFetch — if you prefer tighter controls, restrict network access or require explicit user confirmation before each check. 4) If the scanner returns UNSCANNED, CAUTION, or FAILED, follow your normal manual review processes (review source code, audit permissions) rather than relying solely on the automated verdict. The minor instruction inconsistency (fetch raw JSON vs. 'no raw JSON' reporting rule) is benign but you may want the skill author to clarify that the agent should fetch JSON then format it for users.

Type: OpenClaw Skill Name: pyx-scan Version: 1.1.0 The skill is classified as suspicious due to a critical shell injection vulnerability in `SKILL.md`. User-controlled input (`owner` and `name` from `$ARGUMENTS`) is directly interpolated into a `curl` command without proper sanitization (Step 2). The `allowed-tools: Bash(curl *)` explicitly permits `curl` with arbitrary arguments, allowing an attacker to inject shell metacharacters and execute arbitrary commands on the agent's host system. While the stated purpose of checking skill safety is benign, this implementation flaw presents a significant remote code execution risk.