biome
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides commands to install Biome via official package managers (npm, pnpm, yarn, bun, homebrew, pacman, winget) and by downloading binaries from the project's official GitHub releases. These resources originate from trusted project repositories (e.g.,
github.com/biomejs/biome). - [REMOTE_CODE_EXECUTION]: Documents the process of downloading and executing Biome binaries for different platforms (macOS, Linux, Windows). These installation steps are standard for the tool and target official distribution channels.
- [COMMAND_EXECUTION]: Provides guidance on using the Biome CLI for code formatting, linting, and migration. These operations involve filesystem modifications and integration with version control systems (Git), which are essential features of the linter/formatter.
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the analysis of user-provided source code, which constitutes an indirect prompt injection surface.
- Ingestion points: Project source code files and files specified via CLI arguments are read by the Biome tool.
- Boundary markers: No specific delimiters are defined in the instructions to isolate untrusted code content from the agent's logic.
- Capability inventory: The agent can perform filesystem writes (
--writeflag) and execute git commands as part of the Biome workflow. - Sanitization: The skill does not provide specific instructions for sanitizing embedded code comments or metadata that might contain malicious instructions intended for the agent.
Audit Metadata