a way to standardise communication between AI purposes and exterior instruments or knowledge sources. This standardisation helps to cut back the variety of integrations wanted (from N*M to N+M):
- You need to use community-built MCP servers once you want frequent performance, saving time and avoiding the necessity to reinvent the wheel each time.
- You too can expose your individual instruments and sources, making them obtainable for others to make use of.
In my earlier article, we constructed the analytics toolbox (a set of instruments which may automate your day-to-day routine). We constructed an MCP server and used its capabilities with present shoppers like MCP Inspector or Claude Desktop.
Now, we need to use these instruments immediately in our AI purposes. To do this, let’s construct our personal MCP consumer. We are going to write pretty low-level code, which can even offer you a clearer image of how instruments like Claude Code work together with MCP below the hood.
Moreover, I wish to implement the characteristic that’s presently (July 2025) lacking from Claude Desktop: the power for the LLM to routinely examine whether or not it has an appropriate immediate template for the duty at hand and use it. Proper now, it’s important to decide the template manually, which isn’t very handy.
As a bonus, I can even share a high-level implementation utilizing the smolagents framework, which is good for eventualities once you work solely with MCP instruments and don’t want a lot customisation.
MCP protocol overview
Right here’s a fast recap of the MCP to make sure we’re on the identical web page. MCP is a protocol developed by Anthropic to standardise the way in which LLMs work together with the surface world.
It follows a client-server structure and consists of three major parts:
- Host is the user-facing utility.
- MCP consumer is a part throughout the host that establishes a one-to-one reference to the server and communicates utilizing messages outlined by the MCP protocol.
- MCP server exposes capabilities equivalent to immediate templates, sources and instruments.
Since we’ve already carried out the MCP server earlier than, this time we’ll give attention to constructing the MCP consumer. We are going to begin with a comparatively easy implementation and later add the power to dynamically choose immediate templates on the fly.
You will discover the complete code on GitHub.
Constructing the MCP chatbot
Let’s start with the preliminary setup: we’ll load the Anthropic API key from a config file and regulate Python’s asyncio occasion loop to help nested occasion loops.
# Load configuration and surroundings
with open('../../config.json') as f:
config = json.load(f)
os.environ["ANTHROPIC_API_KEY"] = config['ANTHROPIC_API_KEY']
nest_asyncio.apply()Let’s begin by constructing a skeleton of our program to get a transparent image of the applying’s high-level structure.
async def major():
"""Principal entry level for the MCP ChatBot utility."""
chatbot = MCP_ChatBot()
strive:
await chatbot.connect_to_servers()
await chatbot.chat_loop()
lastly:
await chatbot.cleanup()
if __name__ == "__main__":
asyncio.run(major())We begin by creating an occasion of the MCP_ChatBot class. The chatbot begins by discovering obtainable MCP capabilities (iterating via all configured MCP servers, establishing connections and requesting their lists of capabilities).
As soon as connections are arrange, we’ll initialise an infinite loop the place the chatbot listens to the person queries, calls instruments when wanted and continues this cycle till the method is stopped manually.
Lastly, we’ll carry out a cleanup step to shut all open connections.
Let’s now stroll via every stage in additional element.
Initialising the ChatBot class
Let’s begin by creating the category and defining the __init__ methodology. The primary fields of the ChatBot class are:
exit_stackmanages the lifecycle of a number of async threads (connections to MCP servers), guaranteeing that each one connections will likely be closed appropriately, even when we face an error throughout execution. This logic is carried out within thecleanupperform.anthropicis a consumer for Anthropic API used to ship messages to LLM.available_toolsandavailable_promptsare the lists of instruments and prompts uncovered by all MCP servers we’re related to.classesis a mapping of instruments, prompts and sources to their respective MCP classes. This enables the chatbot to route requests to the right MCP server when the LLM selects a selected device.
class MCP_ChatBot:
"""
MCP (Mannequin Context Protocol) ChatBot that connects to a number of MCP servers
and offers a conversational interface utilizing Anthropic's Claude.
Helps instruments, prompts, and sources from related MCP servers.
"""
def __init__(self):
self.exit_stack = AsyncExitStack()
self.anthropic = Anthropic() # Shopper for Anthropic API
self.available_tools = [] # Instruments from all related servers
self.available_prompts = [] # Prompts from all related servers
self.classes = {} # Maps device/immediate/useful resource names to MCP classes
async def cleanup(self):
"""Clear up sources and shut all connections."""
await self.exit_stack.aclose()Connecting to servers
The primary job for our chatbot is to provoke connections with all configured MCP servers and uncover what capabilities we are able to use.
The record of MCP servers that our agent can connect with is outlined within the server_config.json file. I’ve arrange connections with three MCP servers:
- analyst_toolkit is my implementation of the on a regular basis analytical instruments we mentioned within the earlier article,
- Filesystem permits the agent to work with recordsdata,
- Fetch helps LLMs retrieve the content material of webpages and convert it from HTML to markdown for higher readability.
{
"mcpServers": {
"analyst_toolkit": {
"command": "uv",
"args": [
"--directory",
"/path/to/github/mcp-analyst-toolkit/src/mcp_server",
"run",
"server.py"
],
"env": {
"GITHUB_TOKEN": "your_github_token"
}
},
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/Users/marie/Desktop",
"/Users/marie/Documents/github"
]
},
"fetch": {
"command": "uvx",
"args": ["mcp-server-fetch"]
}
}
}First, we’ll learn the config file, parse it after which join to every listed server.
async def connect_to_servers(self):
"""Load server configuration and connect with all configured MCP servers."""
strive:
with open("server_config.json", "r") as file:
knowledge = json.load(file)
servers = knowledge.get("mcpServers", {})
for server_name, server_config in servers.gadgets():
await self.connect_to_server(server_name, server_config)
besides Exception as e:
print(f"Error loading server config: {e}")
traceback.print_exc()
increaseFor every server, we carry out a number of steps to determine the connection:
- On the transport stage, we launch the MCP server as a stdio course of and get streams for sending and receiving messages.
- On the session stage, we create a
ClientSessionincorporating the streams, after which we carry out the MCP handshake by callinginitializemethodology. - We registered each the session and transport objects within the context supervisor
exit_stackto make sure that all connections will likely be closed correctly ultimately. - The final step is to register server capabilities. We wrapped this performance right into a separate perform, and we’ll talk about it shortly.
async def connect_to_server(self, server_name, server_config):
"""Connect with a single MCP server and register its capabilities."""
strive:
server_params = StdioServerParameters(**server_config)
stdio_transport = await self.exit_stack.enter_async_context(
stdio_client(server_params)
)
learn, write = stdio_transport
session = await self.exit_stack.enter_async_context(
ClientSession(learn, write)
)
await session.initialize()
await self._register_server_capabilities(session, server_name)
besides Exception as e:
print(f"Error connecting to {server_name}: {e}")
traceback.print_exc()Registering capabilities entails iterating over all of the instruments, prompts and sources retrieved from the session. In consequence, we replace the inner variables classes (mapping between sources and a selected session between the MCP consumer and server), available_prompts and available_tools.
async def _register_server_capabilities(self, session, server_name):
"""Register instruments, prompts and sources from a single server."""
capabilities = [
("tools", session.list_tools, self._register_tools),
("prompts", session.list_prompts, self._register_prompts),
("resources", session.list_resources, self._register_resources)
]
for capability_name, list_method, register_method in capabilities:
strive:
response = await list_method()
await register_method(response, session)
besides Exception as e:
print(f"Server {server_name} would not help {capability_name}: {e}")
async def _register_tools(self, response, session):
"""Register instruments from server response."""
for device in response.instruments:
self.classes[tool.name] = session
self.available_tools.append({
"title": device.title,
"description": device.description,
"input_schema": device.inputSchema
})
async def _register_prompts(self, response, session):
"""Register prompts from server response."""
if response and response.prompts:
for immediate in response.prompts:
self.classes[prompt.name] = session
self.available_prompts.append({
"title": immediate.title,
"description": immediate.description,
"arguments": immediate.arguments
})
async def _register_resources(self, response, session):
"""Register sources from server response."""
if response and response.sources:
for useful resource in response.sources:
resource_uri = str(useful resource.uri)
self.classes[resource_uri] = sessionBy the top of this stage, our MCP_ChatBot object has the whole lot it wants to begin interacting with customers:
- connections to all configured MCP servers are established,
- all prompts, sources and instruments are registered, together with descriptions wanted for LLM to know find out how to use these capabilities,
- mappings between these sources and their respective classes are saved, so we all know precisely the place to ship every request.
Chat loop
So, it’s time to begin our chat with customers by creating the chat_loop perform.
We are going to first share all of the obtainable instructions with the person:
- itemizing sources, instruments and prompts
- executing a device name
- viewing a useful resource
- utilizing a immediate template
- quitting the chat (it’s necessary to have a transparent technique to exit the infinite loop).
After that, we’ll enter an infinite loop the place, based mostly on person enter, we’ll execute the suitable motion: whether or not it’s one of many instructions above or making a request to the LLM.
async def chat_loop(self):
"""Principal interactive chat loop with command processing."""
print("nMCP Chatbot Began!")
print("Instructions:")
print(" give up - Exit the chatbot")
print(" @durations - Present obtainable changelog durations")
print(" @ - View changelog for particular interval")
print(" /instruments - Record obtainable instruments")
print(" /device - Execute a device with arguments")
print(" /prompts - Record obtainable prompts")
print(" /immediate - Execute a immediate with arguments")
whereas True:
strive:
question = enter("nQuery: ").strip()
if not question:
proceed
if question.decrease() == 'give up':
break
# Deal with useful resource requests (@command)
if question.startswith('@'):
interval = question[1:]
resource_uri = "changelog://durations" if interval == "durations" else f"changelog://{interval}"
await self.get_resource(resource_uri)
proceed
# Deal with slash instructions
if question.startswith('/'):
elements = self._parse_command_arguments(question)
if not elements:
proceed
command = elements[0].decrease()
if command == '/instruments':
await self.list_tools()
elif command == '/device':
if len(elements) < 2:
print("Utilization: /device ")
proceed
tool_name = elements[1]
args = self._parse_prompt_arguments(elements[2:])
await self.execute_tool(tool_name, args)
elif command == '/prompts':
await self.list_prompts()
elif command == '/immediate':
if len(elements) < 2:
print("Utilization: /immediate ")
proceed
prompt_name = elements[1]
args = self._parse_prompt_arguments(elements[2:])
await self.execute_prompt(prompt_name, args)
else:
print(f"Unknown command: {command}")
proceed
# Course of common queries
await self.process_query(question)
besides Exception as e:
print(f"nError in chat loop: {e}")
traceback.print_exc() There are a bunch of helper features to parse arguments and return the lists of accessible instruments and prompts we registered earlier. Because it’s pretty easy, I gained’t go into a lot element right here. You’ll be able to examine the complete code in case you are .
As an alternative, let’s dive deeper into how the interactions between the MCP consumer and server work in several eventualities.
When working with sources, we use the self.classes mapping to seek out the suitable session (with a fallback choice if wanted) after which use that session to learn the useful resource.
async def get_resource(self, resource_uri):
"""Retrieve and show content material from an MCP useful resource."""
session = self.classes.get(resource_uri)
# Fallback: discover any session that handles this useful resource sort
if not session and resource_uri.startswith("changelog://"):
session = subsequent(
(sess for uri, sess in self.classes.gadgets()
if uri.startswith("changelog://")),
None
)
if not session:
print(f"Useful resource '{resource_uri}' not discovered.")
return
strive:
consequence = await session.read_resource(uri=resource_uri)
if consequence and consequence.contents:
print(f"nResource: {resource_uri}")
print("Content material:")
print(consequence.contents[0].textual content)
else:
print("No content material obtainable.")
besides Exception as e:
print(f"Error studying useful resource: {e}")
traceback.print_exc()To execute a device, we comply with the same course of: begin by discovering the session after which use it to name the device, passing its title and arguments.
async def execute_tool(self, tool_name, args):
"""Execute an MCP device immediately with given arguments."""
session = self.classes.get(tool_name)
if not session:
print(f"Instrument '{tool_name}' not discovered.")
return
strive:
consequence = await session.call_tool(tool_name, arguments=args)
print(f"nTool '{tool_name}' consequence:")
print(consequence.content material)
besides Exception as e:
print(f"Error executing device: {e}")
traceback.print_exc()No shock right here. The identical strategy works for executing the immediate.
async def execute_prompt(self, prompt_name, args):
"""Execute an MCP immediate with given arguments and course of the consequence."""
session = self.classes.get(prompt_name)
if not session:
print(f"Immediate '{prompt_name}' not discovered.")
return
strive:
consequence = await session.get_prompt(prompt_name, arguments=args)
if consequence and consequence.messages:
prompt_content = consequence.messages[0].content material
textual content = self._extract_prompt_text(prompt_content)
print(f"nExecuting immediate '{prompt_name}'...")
await self.process_query(textual content)
besides Exception as e:
print(f"Error executing immediate: {e}")
traceback.print_exc()The one main use case we haven’t coated but is dealing with a basic, free-form enter from a person (not certainly one of particular instructions).
On this case, we ship the preliminary request to the LLM first, then we parse the output, defining whether or not there are any device calls. If device calls are current, we execute them. In any other case, we exit the infinite loop and return the reply to the person.
async def process_query(self, question):
"""Course of a person question via Anthropic's Claude, dealing with device calls iteratively."""
messages = [{'role': 'user', 'content': query}]
whereas True:
response = self.anthropic.messages.create(
max_tokens=2024,
mannequin='claude-3-7-sonnet-20250219',
instruments=self.available_tools,
messages=messages
)
assistant_content = []
has_tool_use = False
for content material in response.content material:
if content material.sort == 'textual content':
print(content material.textual content)
assistant_content.append(content material)
elif content material.sort == 'tool_use':
has_tool_use = True
assistant_content.append(content material)
messages.append({'position': 'assistant', 'content material': assistant_content})
# Execute the device name
session = self.classes.get(content material.title)
if not session:
print(f"Instrument '{content material.title}' not discovered.")
break
consequence = await session.call_tool(content material.title, arguments=content material.enter)
messages.append({
"position": "person",
"content material": [{
"type": "tool_result",
"tool_use_id": content.id,
"content": result.content
}]
})
if not has_tool_use:
breakSo, we’ve now totally coated how the MCP chatbot truly works below the hood. Now, it’s time to check it in motion. You’ll be able to run it from the command line interface with the next command.
python mcp_client_example_base.pyOnce you run the chatbot, you’ll first see the next introduction message outlining potential choices:
MCP Chatbot Began!
Instructions:
give up - Exit the chatbot
@durations - Present obtainable changelog durations
@ - View changelog for particular interval
/instruments - Record obtainable instruments
/device - Execute a device with arguments
/prompts - Record obtainable prompts
/immediate - Execute a immediate with arguments From there, you may check out completely different instructions, for instance,
- name the device to record the databases obtainable within the DB
- record all obtainable prompts
- use the immediate template, calling it like this
/immediate sql_query_prompt query=”What number of prospects did we've in Might 2024?”.
Lastly, I can end your chat by typing give up.
Question: /device list_databases
[07/02/25 18:27:28] INFO Processing request of sort CallToolRequest server.py:619
Instrument 'list_databases' consequence:
[TextContent(type='text', text='INFORMATION_SCHEMAndatasetsndefaultnecommercenecommerce_dbninformation_schemansystemn', annotations=None, meta=None)]
Question: /prompts
Out there prompts:
- sql_query_prompt: Create a SQL question immediate
Arguments:
- query
Question: /immediate sql_query_prompt query="What number of prospects did we've in Might 2024?"
[07/02/25 18:28:21] INFO Processing request of sort GetPromptRequest server.py:619
Executing immediate 'sql_query_prompt'...
I am going to create a SQL question to seek out the variety of prospects in Might 2024.
[07/02/25 18:28:25] INFO Processing request of sort CallToolRequest server.py:619
Based mostly on the question outcomes, here is the ultimate SQL question:
```sql
choose uniqExact(user_id) as customer_count
from ecommerce.classes
the place toStartOfMonth(action_date) = '2024-05-01'
format TabSeparatedWithNames
```
Question: /device execute_sql_query question="choose uniqExact(user_id) as customer_count from ecommerce.classes the place toStartOfMonth(action_date) = '2024-05-01' format TabSeparatedWithNames"
I am going to enable you to execute this SQL question to get the distinctive buyer depend for Might 2024. Let me run this for you.
[07/02/25 18:30:09] INFO Processing request of sort CallToolRequest server.py:619
The question has been executed efficiently. The outcomes present that there have been 246,852 distinctive prospects (distinctive user_ids) in Might 2024 based mostly on the ecommerce.classes desk.
Question: give upAppears to be like fairly cool! Our fundamental model is working nicely! Now, it’s time to take it one step additional and make our chatbot smarter by educating it to counsel related prompts on the fly based mostly on buyer enter.
Immediate solutions
In apply, suggesting immediate templates that greatest match the person’s job might be extremely useful. Proper now, customers of our chatbot have to both already learn about obtainable prompts or not less than be curious sufficient to discover them on their very own to learn from what we’ve constructed. By including a immediate solutions characteristic, we are able to do that discovery for our customers and make our chatbot considerably extra handy and user-friendly.
Let’s brainstorm methods so as to add this performance. I might strategy this characteristic within the following approach:
Consider the relevance of the prompts utilizing the LLM. Iterate via all obtainable immediate templates and, for every one, assess whether or not the immediate is an effective match for the person’s question.
Counsel an identical immediate to the person. If we discovered the related immediate template, share it with the person and ask whether or not they wish to execute it.
Merge the immediate template with the person enter. If the person accepts, mix the chosen immediate with the unique question. Since immediate templates have placeholders, we’d want the LLM to fill them in. As soon as we’ve merged the immediate template with the person’s question, we’ll have an up to date message able to ship to the LLM.
We are going to add this logic to the process_query perform. Because of our modular design, it’s fairly simple so as to add this enhancement with out disrupting the remainder of the code.
Let’s begin by implementing a perform to seek out essentially the most related immediate template. We are going to use the LLM to guage every immediate and assign it a relevance rating from 0 to five. After that, we’ll filter out any prompts with a rating of two or decrease and return solely essentially the most related one (the one with the best relevance rating among the many remaining outcomes).
async def _find_matching_prompt(self, question):
"""Discover a matching immediate for the given question utilizing LLM analysis."""
if not self.available_prompts:
return None
# Use LLM to guage immediate relevance
prompt_scores = []
for immediate in self.available_prompts:
# Create analysis immediate for the LLM
evaluation_prompt = f"""
You might be an professional at evaluating whether or not a immediate template is related for a person question.
Consumer Question: "{question}"
Immediate Template:
- Title: {immediate['name']}
- Description: {immediate['description']}
Charge the relevance of this immediate template for the person question on a scale of 0-5:
- 0: Utterly irrelevant
- 1: Barely related
- 2: Considerably related
- 3: Reasonably related
- 4: Extremely related
- 5: Excellent match
Think about:
- Does the immediate template tackle the person's intent?
- Would utilizing this immediate template present a greater response than a generic question?
- Are the subjects and context aligned?
Reply with solely a single quantity (0-5) and no different textual content.
"""
strive:
response = self.anthropic.messages.create(
max_tokens=10,
mannequin='claude-3-7-sonnet-20250219',
messages=[{'role': 'user', 'content': evaluation_prompt}]
)
# Extract the rating from the response
score_text = response.content material[0].textual content.strip()
rating = int(score_text)
if rating >= 3: # Solely think about prompts with rating >= 3
prompt_scores.append((immediate, rating))
besides Exception as e:
print(f"Error evaluating immediate {immediate['name']}: {e}")
proceed
# Return the immediate with the best rating
if prompt_scores:
best_prompt, best_score = max(prompt_scores, key=lambda x: x[1])
return best_prompt
return NoneThe subsequent perform we have to implement is one that mixes the chosen immediate template with the person enter. We are going to depend on the LLM to intelligently mix them, filling all placeholders as wanted.
async def _combine_prompt_with_query(self, prompt_name, user_query):
"""Use LLM to mix immediate template with person question."""
# First, get the immediate template content material
session = self.classes.get(prompt_name)
if not session:
print(f"Immediate '{prompt_name}' not discovered.")
return None
strive:
# Discover the immediate definition to get its arguments
prompt_def = None
for immediate in self.available_prompts:
if immediate['name'] == prompt_name:
prompt_def = immediate
break
# Put together arguments for the immediate template
args = {}
if prompt_def and prompt_def.get('arguments'):
for arg in prompt_def['arguments']:
arg_name = arg.title if hasattr(arg, 'title') else arg.get('title', '')
if arg_name:
# Use placeholder format for arguments
args[arg_name] = '<' + str(arg_name) + '>'
# Get the immediate template with arguments
consequence = await session.get_prompt(prompt_name, arguments=args)
if not consequence or not consequence.messages:
print(f"Couldn't retrieve immediate template for '{prompt_name}'")
return None
prompt_content = consequence.messages[0].content material
prompt_text = self._extract_prompt_text(prompt_content)
# Create mixture immediate for the LLM
combination_prompt = f"""
You might be an professional at combining immediate templates with person queries to create optimized prompts.
Authentic Consumer Question: "{user_query}"
Immediate Template:
{prompt_text}
Your job:
1. Analyze the person's question and the immediate template
2. Mix them intelligently to create a single, coherent immediate
3. Make sure the person's particular query/request is addressed throughout the context of the template
4. Keep the construction and intent of the template whereas incorporating the person's question
Reply with solely the mixed immediate textual content, no explanations or further textual content.
"""
response = self.anthropic.messages.create(
max_tokens=2048,
mannequin='claude-3-7-sonnet-20250219',
messages=[{'role': 'user', 'content': combination_prompt}]
)
return response.content material[0].textual content.strip()
besides Exception as e:
print(f"Error combining immediate with question: {e}")
return NoneThen, we’ll merely replace the process_query logic to examine for matching prompts, ask the person for affirmation and resolve which message to ship to the LLM.
async def process_query(self, question):
"""Course of a person question via Anthropic's Claude, dealing with device calls iteratively."""
# Test if there is a matching immediate first
matching_prompt = await self._find_matching_prompt(question)
if matching_prompt:
print(f"Discovered matching immediate: {matching_prompt['name']}")
print(f"Description: {matching_prompt['description']}")
# Ask person in the event that they need to use the immediate template
use_prompt = enter("Would you want to make use of this immediate template? (y/n): ").strip().decrease()
if use_prompt == 'y' or use_prompt == 'sure':
print("Combining immediate template together with your question...")
# Use LLM to mix immediate template with person question
combined_prompt = await self._combine_prompt_with_query(matching_prompt['name'], question)
if combined_prompt:
print(f"Mixed immediate created. Processing...")
# Course of the mixed immediate as a substitute of the unique question
messages = [{'role': 'user', 'content': combined_prompt}]
else:
print("Failed to mix immediate template. Utilizing authentic question.")
messages = [{'role': 'user', 'content': query}]
else:
# Use authentic question if person would not need to use the immediate
messages = [{'role': 'user', 'content': query}]
else:
# Course of the unique question if no matching immediate discovered
messages = [{'role': 'user', 'content': query}]
# print(messages)
# Course of the ultimate question (both authentic or mixed)
whereas True:
response = self.anthropic.messages.create(
max_tokens=2024,
mannequin='claude-3-7-sonnet-20250219',
instruments=self.available_tools,
messages=messages
)
assistant_content = []
has_tool_use = False
for content material in response.content material:
if content material.sort == 'textual content':
print(content material.textual content)
assistant_content.append(content material)
elif content material.sort == 'tool_use':
has_tool_use = True
assistant_content.append(content material)
messages.append({'position': 'assistant', 'content material': assistant_content})
# Log device name data
print(f"n[TOOL CALL] Instrument: {content material.title}")
print(f"[TOOL CALL] Arguments: {json.dumps(content material.enter, indent=2)}")
# Execute the device name
session = self.classes.get(content material.title)
if not session:
print(f"Instrument '{content material.title}' not discovered.")
break
consequence = await session.call_tool(content material.title, arguments=content material.enter)
# Log device consequence
print(f"[TOOL RESULT] Instrument: {content material.title}")
print(f"[TOOL RESULT] Content material: {consequence.content material}")
messages.append({
"position": "person",
"content material": [{
"type": "tool_result",
"tool_use_id": content.id,
"content": result.content
}]
})
if not has_tool_use:
breakNow, let’s check our up to date model with a query about our knowledge. Excitingly, the chatbot was capable of finding the suitable immediate and use it to seek out the suitable reply.
Question: What number of prospects did we've in Might 2024?
Discovered matching immediate: sql_query_prompt
Description: Create a SQL question immediate
Would you want to make use of this immediate template? (y/n): y
Combining immediate template together with your question...
[07/05/25 14:38:58] INFO Processing request of sort GetPromptRequest server.py:619
Mixed immediate created. Processing...
I am going to write a question to depend distinctive prospects who had classes in Might 2024. Since this can be a enterprise metric, I am going to exclude fraudulent classes.
[TOOL CALL] Instrument: execute_sql_query
[TOOL CALL] Arguments: {
"question": "/* Rely distinct customers with non-fraudulent classes in Might 2024n Utilizing uniqExact for exact person countn Filtering for Might 2024 utilizing toStartOfMonth and including date vary */nSELECT n uniqExactIf(s.user_id, s.is_fraud = 0) AS active_customers_countnFROM ecommerce.classes snWHERE toStartOfMonth(action_date) = toDate('2024-05-01')nFORMAT TabSeparatedWithNames"
}
[07/05/25 14:39:17] INFO Processing request of sort CallToolRequest server.py:619
[TOOL RESULT] Instrument: execute_sql_query
[TOOL RESULT] Content material: [TextContent(type='text', text='active_customers_countn245287n', annotations=None, meta=None)]
The question exhibits we had 245,287 distinctive prospects with professional (non-fraudulent) classes in Might 2024. This is a breakdown of why I wrote the question this fashion:
1. Used uniqExactIf() to get exact depend of distinctive customers whereas excluding fraudulent classes in a single step
2. Used toStartOfMonth() to make sure we seize all days in Might 2024
3. Specified the date format correctly with toDate('2024-05-01')
4. Used TabSeparatedWithNames format as required
5. Offered a significant column alias
Would you wish to see any variations of this evaluation, equivalent to together with fraudulent classes or breaking down the numbers by nation?It’s all the time a good suggestion to check damaging examples as nicely. On this case, the chatbot behaves as anticipated and doesn’t counsel an SQL-related immediate when given an unrelated query.
Question: How are you?
I ought to notice that I am an AI assistant centered on serving to you're employed with the obtainable instruments, which embrace executing SQL queries, getting database/desk data, and accessing GitHub PR knowledge. I haven't got a device particularly for responding to private questions.
I might help you:
- Question a ClickHouse database
- Record databases and describe tables
- Get details about GitHub Pull Requests
What would you wish to learn about these areas?Now that our chatbot is up and operating, we’re able to wrap issues up.
BONUS: fast and simple MCP consumer with smolagents
We’ve checked out low-level code that permits constructing extremely customised MCP shoppers, however many use instances require solely fundamental performance. So, I made a decision to share with you a fast and simple implementation for eventualities once you want simply the instruments. We are going to use certainly one of my favorite agent frameworks — smolagents from HuggingFace (I’ve mentioned this framework intimately in my earlier article).
# wanted imports
from smolagents import CodeAgent, DuckDuckGoSearchTool, LiteLLMModel, VisitWebpageTool, ToolCallingAgent, ToolCollection
from mcp import StdioServerParameters
import json
import os
# setting OpenAI APIKey
with open('../../config.json') as f:
config = json.hundreds(f.learn())
os.environ["OPENAI_API_KEY"] = config['OPENAI_API_KEY']
# defining the LLM
mannequin = LiteLLMModel(
model_id="openai/gpt-4o-mini",
max_tokens=2048
)
# configuration for the MCP server
server_parameters = StdioServerParameters(
command="uv",
args=[
"--directory",
"/path/to/github/mcp-analyst-toolkit/src/mcp_server",
"run",
"server.py"
],
env={"GITHUB_TOKEN": "github_"},
)
# immediate
CLICKHOUSE_PROMPT_TEMPLATE = """
You're a senior knowledge analyst with greater than 10 years of expertise writing advanced SQL queries, particularly optimized for ClickHouse to reply person questions.
## Database Schema
You might be working with an e-commerce analytics database containing the next tables:
### Desk: ecommerce.customers
**Description:** Buyer data for the net store
**Main Key:** user_id
**Fields:**
- user_id (Int64) - Distinctive buyer identifier (e.g., 1000004, 3000004)
- nation (String) - Buyer's nation of residence (e.g., "Netherlands", "United Kingdom")
- is_active (Int8) - Buyer standing: 1 = energetic, 0 = inactive
- age (Int32) - Buyer age in full years (e.g., 31, 72)
### Desk: ecommerce.classes
**Description:** Consumer session knowledge and transaction data
**Main Key:** session_id
**International Key:** user_id (references ecommerce.customers.user_id)
**Fields:**
- user_id (Int64) - Buyer identifier linking to customers desk (e.g., 1000004, 3000004)
- session_id (Int64) - Distinctive session identifier (e.g., 106, 1023)
- action_date (Date) - Session begin date (e.g., "2021-01-03", "2024-12-02")
- session_duration (Int32) - Session length in seconds (e.g., 125, 49)
- os (String) - Working system used (e.g., "Home windows", "Android", "iOS", "MacOS")
- browser (String) - Browser used (e.g., "Chrome", "Safari", "Firefox", "Edge")
- is_fraud (Int8) - Fraud indicator: 1 = fraudulent session, 0 = professional
- income (Float64) - Buy quantity in USD (0.0 for non-purchase classes, >0 for purchases)
## ClickHouse-Particular Pointers
1. **Use ClickHouse-optimized features:**
- uniqExact() for exact distinctive counts
- uniqExactIf() for conditional distinctive counts
- quantile() features for percentiles
- Date features: toStartOfMonth(), toStartOfYear(), in the present day()
2. **Question formatting necessities:**
- All the time finish queries with "format TabSeparatedWithNames"
- Use significant column aliases
- Use correct JOIN syntax when combining tables
- Wrap date literals in quotes (e.g., '2024-01-01')
3. **Efficiency issues:**
- Use acceptable WHERE clauses to filter knowledge
- Think about using HAVING for post-aggregation filtering
- Use LIMIT when discovering high/backside outcomes
4. **Knowledge interpretation:**
- income > 0 signifies a purchase order session
- income = 0 signifies a searching session with out buy
- is_fraud = 1 classes ought to sometimes be excluded from enterprise metrics until particularly analyzing fraud
## Response Format
Present solely the SQL question as your reply. Embrace temporary reasoning in feedback if the question logic is advanced.
## Examples
**Query:** What number of prospects made buy in December 2024?
**Reply:** choose uniqExact(user_id) as prospects from ecommerce.classes the place toStartOfMonth(action_date) = '2024-12-01' and income > 0 format TabSeparatedWithNames
**Query:** What was the fraud fee in 2023, expressed as a proportion?
**Reply:** choose 100 * uniqExactIf(user_id, is_fraud = 1) / uniqExact(user_id) as fraud_rate from ecommerce.classes the place toStartOfYear(action_date) = '2023-01-01' format TabSeparatedWithNames
**Query:** What was the share of customers utilizing Home windows yesterday?
**Reply:** choose 100 * uniqExactIf(user_id, os = 'Home windows') / uniqExact(user_id) as windows_share from ecommerce.classes the place action_date = in the present day() - 1 format TabSeparatedWithNames
**Query:** What was the income from Dutch customers aged 55 and older in December 2024?
**Reply:** choose sum(s.income) as total_revenue from ecommerce.classes as s interior be a part of ecommerce.customers as u on s.user_id = u.user_id the place u.nation = 'Netherlands' and u.age >= 55 and toStartOfMonth(s.action_date) = '2024-12-01' format TabSeparatedWithNames
**Query:** What are the median and interquartile vary (IQR) of buy income for every nation?
**Reply:** choose nation, median(income) as median_revenue, quantile(0.25)(income) as q25_revenue, quantile(0.75)(income) as q75_revenue from ecommerce.classes as s interior be a part of ecommerce.customers as u on u.user_id = s.user_id the place income > 0 group by nation format TabSeparatedWithNames
**Query:** What's the common variety of days between the primary session and the primary buy for customers who made not less than one buy?
**Reply:** choose avg(first_purchase - first_action_date) as avg_days_to_purchase from (choose user_id, min(action_date) as first_action_date, minIf(action_date, income > 0) as first_purchase, max(income) as max_revenue from ecommerce.classes group by user_id) the place max_revenue > 0 format TabSeparatedWithNames
**Query:** What's the variety of classes in December 2024, damaged down by working methods, together with the totals?
**Reply:** choose os, uniqExact(session_id) as session_count from ecommerce.classes the place toStartOfMonth(action_date) = '2024-12-01' group by os with totals format TabSeparatedWithNames
**Query:** Do we've prospects who used a number of browsers throughout 2024? In that case, please calculate the variety of prospects for every mixture of browsers.
**Reply:** choose browsers, depend(*) as customer_count from (choose user_id, arrayStringConcat(arraySort(groupArray(distinct browser)), ', ') as browsers from ecommerce.classes the place toStartOfYear(action_date) = '2024-01-01' group by user_id) group by browsers order by customer_count desc format TabSeparatedWithNames
**Query:** Which browser has the best share of fraud customers?
**Reply:** choose browser, 100 * uniqExactIf(user_id, is_fraud = 1) / uniqExact(user_id) as fraud_rate from ecommerce.classes group by browser order by fraud_rate desc restrict 1 format TabSeparatedWithNames
**Query:** Which nation had the best variety of first-time customers in 2024?
**Reply:** choose nation, depend(distinct user_id) as new_users from (choose user_id, min(action_date) as first_date from ecommerce.classes group by user_id having toStartOfYear(first_date) = '2024-01-01') as t interior be a part of ecommerce.customers as u on t.user_id = u.user_id group by nation order by new_users desc restrict 1 format TabSeparatedWithNames
---
**Your Process:** Utilizing all of the supplied data above, write a ClickHouse SQL question to reply the next buyer query:
{query}
"""
with ToolCollection.from_mcp(server_parameters, trust_remote_code=True) as tool_collection:
agent = ToolCallingAgent(instruments=[*tool_collection.tools], mannequin=mannequin)
immediate = CLICKHOUSE_PROMPT_TEMPLATE.format(
query = 'What number of prospects did we've in Might 2024?'
)
response = agent.run(immediate) In consequence, we acquired the right reply.
If you happen to don’t want a lot customisation or integration with prompts and sources, this implementation is certainly the way in which to go.
Abstract
On this article, we constructed a chatbot that integrates with MCP servers and leverages all the advantages of standardisation to entry instruments, prompts, and sources seamlessly.
We began with a fundamental implementation able to itemizing and accessing MCP capabilities. Then, we enhanced our chatbot with a sensible characteristic that implies related immediate templates to customers based mostly on their enter. This makes our product extra intuitive and user-friendly, particularly for customers unfamiliar with the whole library of accessible prompts.
To implement our chatbot, we used comparatively low-level code, supplying you with a greater understanding of how the MCP protocol works below the hood and what occurs once you use AI instruments like Claude Desktop or Cursor.
As a bonus, we additionally mentioned the smolagents implementation that allows you to shortly deploy an MCP consumer built-in with instruments.
Thanks for studying. I hope this text was insightful. Keep in mind Einstein’s recommendation: “The necessary factor is to not cease questioning. Curiosity has its personal purpose for present.” Might your curiosity lead you to your subsequent nice perception.
Reference
This text is impressed by the “MCP: Construct Wealthy-Context AI Apps with Anthropic” quick course from DeepLearning.AI.
