Skills
Navigate
skills/ortus-boxlang/bx-sites-skills/bx-sites-getting-started/Gen Agent Trust Hub

bx-sites-getting-started

Fail

Audited by Gen Agent Trust Hub on Oct 26, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The instructions describe downloading and executing installation scripts directly from the vendor's domain (https://install.boxlang.io/ and https://install-bvm.boxlang.io/) using curl | bash piped commands.
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute various CLI commands such as bxSites build, bxSites serve, and bxSites new to manage the static site lifecycle and local development server.
  • [EXTERNAL_DOWNLOADS]: The skill fetches software components, dependencies, and management tools from vendor-controlled domains and package managers (install-bx-module, box install).
  • [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection by processing external project data which could contain malicious instructions.
  • Ingestion points: Content is ingested from external paths and archives (GitBook, mkdocs, Notion) via the bxSites migrate command described in SKILL.md.
  • Boundary markers: No explicit boundary markers or instructions to disregard embedded content are defined for the migration or build process.
  • Capability inventory: The skill utilizes shell commands for building and serving content (bxSites build, bxSites serve) in SKILL.md, including a local web server with live reload.
  • Sanitization: No explicit sanitization or validation of the ingested Markdown content is specified during the migration process.
Recommendations
  • HIGH: Downloads and executes remote code from: https://install-bvm.boxlang.io/, https://install.boxlang.io/ - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Oct 26, 2026, 05:02 PM