adhx
Fail
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill documentation explicitly provides a manual installation command that downloads a markdown file from a remote repository and saves it directly into the agent's internal skills directory (
~/.claude/skills/adhx/SKILL.md). This bypasses standard safety checks and allows for arbitrary instruction injection. - [EXTERNAL_DOWNLOADS]: The skill uses
curlto fetch JSON data fromhttps://adhx.com/api/share/tweet/{username}/{statusId}. While the domain is related to the skill's stated purpose, it is not a recognized trusted service and results in processing third-party data. - [COMMAND_EXECUTION]: The workflow involves executing
curlcommands in the terminal to interact with an external API, exposing the agent to potential command injection if parameters are not correctly handled. - [DATA_EXFILTRATION]: While intended to fetch public posts, the use of
curlto send username and status IDs to a third-party service (adhx.com) constitutes a form of external data transmission that should be monitored.
Recommendations
- HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/itsmemeworks/adhx/main/skills/adhx/SKILL.md - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata