saju-fortune
Fail
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to run
npx -y @nomadamas/k-skill@0 instruct saju-fortune. This command fetches and executes code from an untrusted external package registry. The use of the-yflag is dangerous as it automatically accepts the installation and execution of the remote package without user verification. - [PROMPT_INJECTION]: The
SKILL.mdfile directs the agent to follow the output of a remote command as its "primary instructions." This creates a mechanism for remote instruction injection, allowing an external party to override the agent's behavior and safety guidelines. - [EXTERNAL_DOWNLOADS]: The skill requires global installation of the
saju-fortunenpm package. There is no verification of the package's integrity or source, posing a supply chain risk. - [DYNAMIC_EXECUTION]: The skill uses the
nodeCLI to execute dynamically generated JavaScript code blocks via heredoc syntax. This pattern of generating and immediately executing code is high-risk and can lead to code injection if variables are not strictly handled. - [DATA_EXFILTRATION]: The skill is designed to collect highly sensitive personally identifiable information (PII), including the user's full name, exact birth date, birth time, and birth city. Combined with the execution of unverified remote code, this presents a significant risk of data exfiltration to an external server.
Recommendations
- AI detected serious security threats
Audit Metadata