metorial-mcp-session 1.0.6

MCP session management for Metorial

Metorial Python SDK

The official Python SDK for Metorial.

Available Providers

Provider	Import	Format	Description
OpenAI	metorial_openai	OpenAI function calling	GPT-4, GPT-3.5, etc.
Anthropic	metorial_anthropic	Claude tool format	Claude 3.5, Claude 3, etc.
Google	metorial_google	Gemini function declarations	Gemini Pro, Gemini Flash
Mistral	metorial_mistral	Mistral function calling	Mistral Large, Codestral
DeepSeek	metorial_deepseek	OpenAI-compatible	DeepSeek Chat, DeepSeek Coder
TogetherAI	metorial_togetherai	OpenAI-compatible	Llama, Mixtral, etc.
XAI	metorial_xai	OpenAI-compatible	Grok models
AI SDK	metorial_ai_sdk	Framework tools	Vercel AI SDK, etc.

Installation

# Install core metorial package (includes all provider adapters)
pip install metorial

# Install with specific providers (includes provider client libraries)
pip install metorial[openai,anthropic,google,mistral,deepseek,togetherai,xai]

# Or install individual providers
pip install metorial[openai]      # Includes openai client
pip install metorial[anthropic]   # Includes anthropic client
# ... etc

Quick Start

import asyncio
from metorial import Metorial
from openai import AsyncOpenAI

async def main():
  metorial = Metorial(api_key="your-metorial-api-key")
  openai_client = AsyncOpenAI(api_key="your-openai-api-key")
  
  response = await metorial.run(
    "What are the latest commits in the metorial/websocket-explorer repository?",
    "your-server-deployment-id",  # can also be a list
    openai_client,
    model="gpt-4o",
    max_iterations=25
  )
  
  print("Response:", response)

asyncio.run(main())

That's it! metorial.run() automatically:

• Creates a session with your MCP server
• Formats tools for your AI provider
• Handles the execution loop
• Manages tool execution
• Returns the final response

Synchronous Usage

For synchronous applications, use MetorialSync:

from metorial import MetorialSync
from openai import OpenAI

metorial = MetorialSync(api_key="your-metorial-api-key")
openai_client = OpenAI(api_key="your-openai-api-key")

response = metorial.run(
  "What are the latest commits in the metorial/websocket-explorer repository?",
  "your-server-deployment-id",  # can also be a list
  openai_client,
  model="gpt-4o",
  max_iterations=25
)

print("Response:", response)

Provider Examples

Metorial works with all major AI providers. Here are examples using metorial.run():

OpenAI (GPT-4, GPT-3.5)

from metorial import Metorial
from openai import AsyncOpenAI

metorial = Metorial(api_key="your-metorial-api-key")
openai_client = AsyncOpenAI(api_key="your-openai-api-key")

response = await metorial.run(
  "What are the latest commits?",
  "your-deployment-id",
  openai_client,
  model="gpt-4o"
)

Anthropic (Claude)

from metorial import Metorial
import anthropic

metorial = Metorial(api_key="your-metorial-api-key")
anthropic_client = anthropic.AsyncAnthropic(api_key="your-anthropic-api-key")

response = await metorial.run(
  "What are the latest commits?",
  "your-deployment-id", 
  anthropic_client,
  model="claude-3-5-sonnet-20241022"
)

Google (Gemini)

from metorial import Metorial
import google.generativeai as genai

metorial = Metorial(api_key="your-metorial-api-key")
genai.configure(api_key="your-google-api-key")
google_client = genai.GenerativeModel('gemini-pro')

response = await metorial.run(
  "What are the latest commits?",
  "your-deployment-id",
  google_client,
  model="gemini-pro"
)

Mistral AI

from metorial import Metorial
from mistralai import AsyncMistral

metorial = Metorial(api_key="your-metorial-api-key")
mistral_client = AsyncMistral(api_key="your-mistral-api-key")

response = await metorial.run(
  "What are the latest commits?",
  "your-deployment-id",
  mistral_client,
  model="mistral-large-latest"
)

DeepSeek

from metorial import Metorial
from openai import AsyncOpenAI

metorial = Metorial(api_key="your-metorial-api-key")
deepseek_client = AsyncOpenAI(
  api_key="your-deepseek-api-key",
  base_url="https://api.deepseek.com"
)

response = await metorial.run(
  "What are the latest commits?",
  "your-deployment-id",
  deepseek_client,
  model="deepseek-chat"
)

Together AI

from metorial import Metorial
from openai import AsyncOpenAI

metorial = Metorial(api_key="your-metorial-api-key")
together_client = AsyncOpenAI(
  api_key="your-together-api-key",
  base_url="https://api.together.xyz/v1"
)

response = await metorial.run(
  "What are the latest commits?",
  "your-deployment-id",
  together_client,
  model="meta-llama/Llama-2-70b-chat-hf"
)

XAI (Grok)

from metorial import Metorial
from openai import AsyncOpenAI

metorial = Metorial(api_key="your-metorial-api-key")
xai_client = AsyncOpenAI(
  api_key="your-xai-api-key",
  base_url="https://api.x.ai/v1"
)

response = await metorial.run(
  "What are the latest commits?",
  "your-deployment-id",
  xai_client,
  model="grok-beta"
)

Advanced Usage

Session Management

For more control over the conversation flow, use session management directly:

import asyncio
from metorial import Metorial, MetorialOpenAI
from openai import AsyncOpenAI

async def main():
  metorial = Metorial(api_key="your-metorial-api-key")
  openai_client = AsyncOpenAI(api_key="your-openai-api-key")
  
  async def session_callback(session):
    messages = [{"role": "user", "content": "What are the latest commits?"}]
    
    for i in range(10):
      # Call OpenAI with Metorial tools
      response = await openai_client.chat.completions.create(
        model="gpt-4o",
        messages=messages,
        tools=session.tools
      )
      
      choice = response.choices[0]
      tool_calls = choice.message.tool_calls
      
      if not tool_calls:
        print(choice.message.content)
        return
      
      # Execute tools through Metorial
      tool_responses = await session.call_tools(tool_calls)
      
      # Add to conversation
      messages.append({
        "role": "assistant",
        "tool_calls": [
          {
            "id": tc.id,
            "type": tc.type,
            "function": {
              "name": tc.function.name,
              "arguments": tc.function.arguments
            }
          } for tc in tool_calls
        ]
      })
      messages.extend(tool_responses)

  await metorial.with_provider_session(
    MetorialOpenAI.chat_completions(openai_client),
    "your-deployment-id",
    session_callback
  )

asyncio.run(main())

Streaming Responses

For real-time streaming responses:

import asyncio
from metorial import Metorial, MetorialOpenAI
from metorial.types import StreamEventType

async def stream_chat():
  metorial = Metorial(api_key="your-metorial-api-key")
  openai_client = AsyncOpenAI(api_key="your-openai-api-key")
  
  async def stream_action(session):
    messages = [{"role": "user", "content": "What are the latest commits?"}]
    
    async for event in metorial.stream(
      openai_client, session, messages, max_iterations=10
    ):
      if event.type == StreamEventType.CONTENT:
        print(f"🤖 {event.content}", end="", flush=True)
      elif event.type == StreamEventType.TOOL_CALL:
        print(f"\n🔧 Executing {len(event.tool_calls)} tool(s)...")
      elif event.type == StreamEventType.COMPLETE:
        print(f"\n✅ Complete! Duration: {event.metadata.get('duration', 0):.2f}s")
      elif event.type == StreamEventType.ERROR:
        print(f"\n❌ Error: {event.error}")
        break
  
  await metorial.with_provider_session(
    MetorialOpenAI.chat_completions(openai_client),
    "your-deployment-id",
    stream_action
  )

asyncio.run(stream_chat())

Batch Processing

Process multiple messages concurrently:

import asyncio

async def batch_example():
  metorial = Metorial(api_key="your-metorial-api-key")
  openai_client = AsyncOpenAI(api_key="your-openai-api-key")
  
  messages = [
    "What are the latest commits?",
    "What are the main features?",
    "How do I get started?"
  ]
  
  results = await metorial.batch_run(
    messages,
    "your-deployment-id",
    openai_client,
    max_iterations=25
  )
  
  for i, result in enumerate(results):
    print(f"Response {i+1}: {result}")

asyncio.run(batch_example())

Error Handling

from metorial import MetorialAPIError

try:
  response = await metorial.run(
    "What are the latest commits?",
    "your-deployment-id",
    openai_client,
    model="gpt-4o"
  )
except MetorialAPIError as e:
  print(f"API Error: {e.message} (Status: {e.status_code})")
except Exception as e:
  print(f"Unexpected error: {e}")

Examples

Check out the examples/ directory for more comprehensive examples.

License

MIT License - see LICENSE file for details.

Support

• 📖 Documentation
• 🐛 GitHub Issues
• 📧 Email Support