For AI & agents
AI & Agent Access
Suger is built to be read by machines as well as people. Assistants, answer engines, and autonomous agents can search the site, read any page as clean Markdown, and call live tools over the Model Context Protocol (MCP).
Search
Every page has instant full-text search. Press ⌘K (or Ctrl K) anywhere, or use the dedicated search page. It runs client-side on a prebuilt index — no query is sent to a server.
Machine-readable content
Clean, presentation-free content for grounding assistants:
/index.md to its URL. /search-index.json The prebuilt MiniSearch index powering on-site and MCP search. Model Context Protocol (MCP)
Connect any MCP client (Claude Desktop, Cursor, IDE agents) to call two read-only tools:
search and fetch_page.
Hosted endpoint (no install)
Streamable HTTP at https://www.suger.io/mcp. Add to your client config:
{
"mcpServers": {
"suger": { "url": "https://www.suger.io/mcp" }
}
} Standalone package (offline / no hosted dependency)
Runs locally over stdio and works against the published content:
{
"mcpServers": {
"suger": {
"command": "npx",
"args": ["-y", "suger-website-mcp"]
}
}
} Discovery manifest: /.well-known/mcp.json.
FAQ
- What can the Suger MCP server do?
- It exposes two read-only tools: search (find pages by keyword) and fetch_page (read a page as clean Markdown). Both cover cloud marketplace GTM, co-sell, private offers, agreements, and billing content on suger.io.
- Do I need an API key?
- No. Everything here is public and read-only. The hosted endpoint and the standalone package both work without authentication.
- How do I read a single page as Markdown?
- Append /index.md to any page URL. For example, https://www.suger.io/platform/cosell/index.md returns the clean Markdown of the Co-Sell page.
- Is the content kept up to date?
- Yes. The search index, per-page Markdown, and llms-full.txt are regenerated automatically on every site build, so they always match the live pages.