parallel-agents
Fail
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: CRITICALREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill performs remote code execution by downloading a script and piping it directly into a shell environment (
curl | sh). This is a critical security risk because the script content is not verified before it is executed on the host system. Evidence:curl -fsSL https://box.ascii.dev/install | shin SKILL.md. - [EXTERNAL_DOWNLOADS]: The skill fetches executable installation scripts from an external, non-whitelisted domain (
box.ascii.dev). - [COMMAND_EXECUTION]: The skill executes multiple shell commands to manage infrastructure and run tasks, including
box login,box new,box fork, andbox ssh. Evidence: Bash script blocks and CLI usage instructions in SKILL.md. - [PROMPT_INJECTION]: The skill establishes a workflow pattern vulnerable to indirect prompt injection. Ingestion points: Workload data and tasks distributed to virtual machines via
run-task.sh. Boundary markers: Absent. Capability inventory: Full command execution on remote machines viabox ssh. Sanitization: Absent.
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