beautiful-mermaid

Warn

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The script scripts/render.ts contains a function ensurePackage that automatically installs the beautiful-mermaid package from NPM at runtime using bun add or npm install if it is not found in the environment.
  • [COMMAND_EXECUTION]: The skill relies on executing shell commands via bun run, npx tsx, and deno run to process diagrams. It also invokes the agent-browser CLI, which executes browser-based operations on local files.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests Mermaid diagram code (via --code or --input) which is then rendered into SVG and eventually HTML.
  • Ingestion points: Mermaid syntax provided by the user or read from files (SKILL.md, scripts/render.ts).
  • Boundary markers: None explicitly present to separate user-provided diagram code from the rendering logic.
  • Capability inventory: File read/write, network access (via package managers), and browser automation.
  • Sanitization: There is no evidence of sanitization for the Mermaid code before it is passed to the rendering library or embedded in the HTML wrapper.
  • [DYNAMIC_EXECUTION]: The scripts/create-html.ts script dynamically generates an HTML file by interpolating the raw content of an SVG file directly into a template, which is then opened in a browser environment via agent-browser.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 15, 2026, 12:20 AM