MCP ServerBETA

The Rise Documentation MCP Server bridges your LLM environment with the Rise knowledge base. It utilizes semantic vector search to retrieve high-fidelity, context-aware documentation chunks, allowing agents to answer technical questions about the Rise Platform with precision.

Coming Soon!
📹 Video Tutorial - Step-by-step setup & usage guide

⚙️ Quick Installation

To connect an MCP-compliant client to this server, add the following to your client's configuration file:

{
  "mcpServers": {
    "rise-docs": {
      "url": "https://mcp.rise.ai/docs",
      "type": "http"
    }
  }
}

🛠️ Available Tools

SearchDocs

Searches the Rise Platform documentation and retrieves coherent, contextual results.

Input Parameters

Parameter	Type	Required	Description
`searchTerm`	string	Yes	The natural language query. For best results, use specific technical terms.

Key Features

🔍Semantic SearchGoes beyond keyword matching using vector-based retrieval to understand the intent behind queries.

📚Deep ContextRetrieves the best matching segment plus the surrounding context (±15 chunks) to ensure the LLM understands the broader topic.

🎯Smart FilteringApplies relevance scoring and thresholding to reduce hallucinations and irrelevant noise.

🏷️Optimized FormattingReturns results with readable source names, titles, and quality indicators optimized for LLM consumption.

Tool Definition

            {
                "name": "SearchDocs",
                "description": "Search Rise documentation and retrieve coherent, contextual results.\n\nRESULT FORMAT:\nEach result includes:\n- Title (if available)\n- Source: Knowledge base name\n- Relevance: Quality indicator (e.g., \"Excellent match ⭐⭐⭐\", \"Very relevant ⭐⭐\", \"Relevant ⭐\", \"Somewhat relevant\", \"Moderate match\")\n- Content: Up to 31 consecutive documentation sections centered around the most relevant match\n\nBEST PRACTICES FOR PRESENTING RESULTS:\n- Present \"Excellent match\" and \"Very relevant\" results confidently to the user\n- For \"Somewhat relevant\" or \"Moderate match\" results, ask clarifying questions\n- Synthesize information from multiple results when they complement each other\n- If results seem loosely related, help the user refine their query\n- Extract key steps, prerequisites, and warnings from the coherent content\n\nENGAGEMENT GUIDELINES:\n- Start with a direct answer from the highest-relevance result\n- Highlight important prerequisites or warnings found in the context\n- Ask follow-up questions when the user's goal or context is unclear\n- Guide users to rephrase vague queries for better results\n- Build on previous conversation context",
                "inputSchema": {
                    "type": "object",
                    "properties": {
                        "searchTerm": {
                            "type": "string",
                            "description": "The search query. Use specific technical terms and include context for best results (e.g., \"Next.js authentication setup\" instead of just \"auth\"). The tool will return relevant documentation sections with surrounding context."
                        }
                    },
                    "required": [
                        "searchTerm"
                    ],
                    "additionalProperties": false,
                    "$schema": "http://json-schema.org/draft-07/schema#"
                }
            }

🔄 Session LifecycleThis server follows the standard MCP protocol. Like any MCP server, it manages stateful sessions to handle tool execution efficiently.

View Session Flow Details

1. Create Session (POST /docs)Client initiates the MCP session and receives a session ID

2. Tool Invocation (MCP Protocol)Client sends tool calls and server executes semantic search

3. Terminate Session (DELETE /docs)Client closes the session and server performs cleanup