building-inferencesh-apps
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill provides instructions to install the platform's CLI and associated development tools by downloading and executing shell scripts from remote servers. These commands target vendor-owned or well-known service domains.
curl -fsSL https://cli.inference.sh | sh(Official platform CLI)curl -LsSf https://astral.sh/uv/install.sh | sh(Astral uv installer)curl -fsSL https://fnm.vercel.app/install | bash(Fast Node Manager via Vercel)curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.0/install.sh | bash(Node Version Manager via GitHub)- [PRIVILEGE_ESCALATION]: The documentation for Windows environments includes instructions to bypass PowerShell execution policies (
-ExecutionPolicy ByPass) to facilitate the automated installation of theuvpackage manager. - [COMMAND_EXECUTION]: The skill relies extensively on the
beltCLI tool for scaffolding, testing, and deploying applications, which involves multiple shell command invocations throughout the development lifecycle. - [INDIRECT_PROMPT_INJECTION]: The skill manages a development surface where the agent implements logic based on local files. It includes defensive instructions to ignore unofficial project structure files (
PROVIDER_STRUCTURE.md) that could mislead the agent, and it promotes the use of schema-based validation (Zod/Pydantic) for application inputs. - [SAFE]: All identified remote code execution and policy bypass patterns are part of the intended primary purpose of the skill and originate from official vendor infrastructure or trusted technology organizations.
Audit Metadata