---

This file is part of Logtalk https://logtalk.org/ SPDX-FileCopyrightText: 1998-2026 Paulo Moura <pmoura@logtalk.org> SPDX-License-Identifier: Apache-2.0

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

birds_mcp

This example demonstrates the mcp_server library by exposing the bird identification expert system (from the birds example) as an MCP (Model Context Protocol) server with elicitation support.

When an MCP client asks to identify a bird, the server uses the elicitation/create MCP method to ask the user questions about bird characteristics (yes/no questions and multiple-choice menus). The answers guide the expert system through the bird taxonomy to identify the species — exactly as the original interactive expert system does, but over the MCP protocol instead of terminal I/O.

Key concepts demonstrated

• Implementing mcp_tool_protocol with the elicitation capability
• Using tool_call/4 with the Elicit closure for interactive tools
• Building JSON Schema for yes/no and enum elicitation requests
• Adapting an existing interactive application for MCP without modifying the original knowledge base

Architecture

The example reuses the bird taxonomy from `examples/birds/` (the order prototype hierarchy (the descriptors category, the birds objects). The expert system's user interaction is reimplemented in birds_mcp.lgt:

• birds_mcp implements mcp_tool_protocol and declares the elicitation capability
• A single tool identify_bird is exposed
• The check/2 predicate threads the Elicit closure through the identification logic
• ask_descriptor/2 dispatches each bird descriptor to either a yes/no elicitation (ask/3) or a multiple-choice menu (menuask/4), based on menu_attribute/2 facts
• Previous answers are memoized in known_/3 to avoid redundant questions (matching the original expert system behavior)

  The server reads/writes JSON-RPC 2.0 messages from/to stdin/stdout as required by MCP.

MCP client configuration

MCP client configuration depends on the MCP client. For example, to add this MCP server to Claude Desktop or VSCode, edit (respectively) the claude_desktop_config.json or mcp.json files and add:

{
        "mcpServers": {
                "my-server": {
                        "command": "swilgt",
                        "args": [
                "-q",
                                "-g", "logtalk_load(birds_mcp(server))",
                                "-t", "halt"
                        ],
                        "env": {
                                "LOGTALKHOME": "/usr/local/share/logtalk",
                                "LOGTALKUSER": "/Users/jdoe/logtalk"
                        }
                }
        }
}

The env definition of the LOGTALKHOME and LOGTALKUSER environment variables may or may not be required (it's usually necessary on macOS). When required, replace the values above with the actual values on your Logtalk setup.

The actual arguments to the integration script (swilgt in the example above) depend on the Prolog backend.

Protocol interaction example

A typical identification session involves the following MCP message exchange:

1. Client sends initialize request
2. Server responds with capabilities including elicitation
3. Client sends tools/list request
4. Server responds with the identify_bird tool
5. Client sends tools/call for identify_bird
6. Server sends elicitation/create requests asking about bird characteristics (e.g., "eats: meat?", "What is the value for size?")
7. Client responds to each elicitation with the user's answer
8. Server returns the identification result Example elicitation request (server -> client):

   {
     "jsonrpc": "2.0",
     "id": "elicit_1",
     "method": "elicitation/create",
     "params": {
       "message": "bill: sharp_hooked?",
       "requestedSchema": {
         "type": "object",
         "properties": {
           "answer": {"type": "string", "enum": ["yes", "no"]}
         },
         "required": ["answer"]
       }
     }
   }

   Client response:

   {
     "jsonrpc": "2.0",
     "id": "elicit_1",
     "result": {
       "action": "accept",
       "content": {"answer": "yes"}
     }
   }