cloud-sandbox
Fail
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the user to install a necessary CLI tool by piping a script from a remote URL (
https://box.ascii.dev/install) directly into the shell (| sh). This execution method bypasses verification and security checks. - [COMMAND_EXECUTION]: The skill uses the
box sshcommand to execute arbitrary shell scripts and commands on a remote virtual machine. This behavior allows for the execution of untrusted logic in a cloud environment. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it is designed to ingest and execute external scripts and repository tests without sanitization.
- Ingestion points: Use of local files like
setup.shand project-level commands such asnpm test(SKILL.md). - Boundary markers: Absent; there are no delimiters or instructions to ignore potential commands within the data being processed.
- Capability inventory: Full shell access and remote process execution via the 'box' CLI (SKILL.md).
- Sanitization: No validation or escaping is applied to the scripts or project content before they are executed on the remote host.
Recommendations
- HIGH: Downloads and executes remote code from: https://box.ascii.dev/install - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata