Overview
What is Fast HTML MCP?
Fast HTML MCP is a Model Context Protocol server for fast HTML generation and editing by AI agents. It uses a five-tier architecture (Assembly–Patch–Read–Raw–Consistency) and provides 15 tools, 22 components, and 25 templates purpose-built for AI-driven page creation with sub-second patch times and token-efficient reading.
How to use Fast HTML MCP?
Install and run with npx -y @aimino/fast-html-mcp-server. Configure the server in your MCP client (Claude Desktop, Cursor, VS Code via GitHub Copilot, or Claude Code) using the provided JSON snippets. Invoke tools via JSON-RPC calls—for example, list_templates, render_page, read_html—following the Ping-Pong Loop: discover, build, inspect, refine, and maintain consistency.
Key features of Fast HTML MCP
- 15 tools across five tiers: Assembly, Patch, Read, Raw, Consistency
- 22 reusable components and 25 pre-compiled templates
- Sub-second patching via #id fast-path (direct string substitution)
- Document Consistency Engine for cross-section data integrity
- Token-efficient read modes: structure, content, compressed, text
- AI-grade token compression and streaming output
- XSS prevention with DOMPurify and atomic writes
Use cases of Fast HTML MCP
- Generate equity research reports with 10+ interdependent sections
- Build and iterate on data dashboards with tables and charts
- Create and patch landing pages, newsletters, and documentation
- Produce financial summaries and invoicing templates
- Maintain consistency across large multi-section documents
FAQ from Fast HTML MCP
What runtime does Fast HTML MCP require?
Node.js 20 or higher, and the server is distributed as an ES module (ESM).
How does Fast HTML MCP achieve sub-second patch speeds?
It uses a #id fast-path for direct string substitution on #id selectors (~10× faster than full AST parse), pre-compiled doT.js templates, no browser runtime (unlike Playwright/Puppeteer), and atomic in-place edits without full DOM re-serialization.
What is the Document Consistency Engine?
It is a dependency-graph-based engine that propagates entity edits across sections, detects stale cross-section references with 100% accuracy, and auto-updates affected sections—achieving ~11 ms worst-case propagation for 100 sections.
How does token compression work for AI agents?
Compression modes reduce HTML size: high mode yields 40–70% reduction, and ai mode produces token-aware minification (~1600 chars for a full report). Read modes also save tokens: text mode strips HTML tags for 97% savings over raw HTML.
What security measures are in Fast HTML MCP?
All output is sanitized with DOMPurify to prevent XSS. File writes are atomic (tmp file + rename) to prevent partial writes.
