browse
Pass
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads the Bun runtime installer from bun.sh, which is a well-known technology service. The process includes a SHA256 checksum verification step to ensure the integrity of the script before execution.
- [PROMPT_INJECTION]: To mitigate risks from untrusted web data, the skill uses '--- BEGIN/END UNTRUSTED EXTERNAL CONTENT ---' boundary markers. It provides explicit instructions to the agent to ignore any commands or instructions found within these markers, satisfying the evidence chain for handling indirect injection.
- [REMOTE_CODE_EXECUTION]: JavaScript execution ('js', 'eval') and browser protocol access ('cdp') are provided for automation. These capabilities are restricted to the browser environment and governed by an internal allowlist to prevent unauthorized system access.
- [COMMAND_EXECUTION]: Local shell commands are used for environment setup and managing project-specific configuration files like CLAUDE.md, which is standard behavior for integrating local development tools.
Audit Metadata