claude-code-best
Fail
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs users to install a global NPM package (
claude-code-best) and clone/run code from an untrusted GitHub repository (github.com/claude-code-best/claude-code). This code claims to be a reverse-engineered fork of a corporate tool, which is a common vector for distributing malicious software. - [REMOTE_CODE_EXECUTION]: Fetches and executes the Bun runtime installation script from
https://bun.sh/installvia pipe to bash. - [DATA_EXFILTRATION]: Encourages the use of a 'remote control' feature that connects the agent to a third-party server hosted on
remote-control.claude-code-best.win. The use of the.winTLD for an unofficial tool is highly suspicious and could be used to exfiltrate session data or credentials. - [EXTERNAL_DOWNLOADS]: The skill documents a plugin system (
--channels) that downloads and executes third-party code from an unspecified 'marketplace'. - [COMMAND_EXECUTION]: The tool described requires and exercises extensive system permissions, including arbitrary shell command execution (
executeShellCommand), file system manipulation, and computer automation (screen control). - [PROMPT_INJECTION]: The skill has a high surface area for indirect prompt injection. It ingests untrusted data from web searches (Bing/Brave) and external plugins while possessing 'CRITICAL' capability tier permissions (file write, subprocess execution, computer control). No boundary markers or sanitization procedures are documented for this data ingestion.
Recommendations
- HIGH: Downloads and executes remote code from: https://bun.sh/install - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata