casual-mcp 0.4.0__py3-none-any.whl → 0.6.0__py3-none-any.whl

casual_mcp-0.4.0.dist-info/METADATA

@@ -1,399 +0,0 @@
-Metadata-Version: 2.4
-Name: casual-mcp
-Version: 0.4.0
-Summary: Multi-server MCP client for LLM tool orchestration
-Author: Alex Stansfield
-License: MIT
-Project-URL: Homepage, https://github.com/AlexStansfield/casual-mcp
-Project-URL: Repository, https://github.com/AlexStansfield/casual-mcp
-Project-URL: Issue Tracker, https://github.com/AlexStansfield/casual-mcp/issues
-Requires-Python: >=3.10
-Description-Content-Type: text/markdown
-License-File: LICENSE
-Requires-Dist: dateparser>=1.2.1
-Requires-Dist: fastapi>=0.115.12
-Requires-Dist: fastmcp>=2.12.4
-Requires-Dist: jinja2>=3.1.6
-Requires-Dist: ollama>=0.4.8
-Requires-Dist: openai>=1.78.0
-Requires-Dist: python-dotenv>=1.1.0
-Requires-Dist: requests>=2.32.3
-Requires-Dist: rich>=14.0.0
-Requires-Dist: typer>=0.19.2
-Requires-Dist: uvicorn>=0.34.2
-Provides-Extra: dev
-Requires-Dist: ruff; extra == "dev"
-Requires-Dist: black; extra == "dev"
-Requires-Dist: mypy; extra == "dev"
-Requires-Dist: pytest; extra == "dev"
-Requires-Dist: coverage; extra == "dev"
-Dynamic: license-file
-
-# 🧠 Casual MCP
-
-![PyPI](https://img.shields.io/pypi/v/casual-mcp)
-![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)
-
-**Casual MCP** is a Python framework for building, evaluating, and serving LLMs with tool-calling capabilities using [Model Context Protocol (MCP)](https://modelcontextprotocol.io).  
-It includes:
-
-- ✅ A multi-server MCP client using [FastMCP](https://github.com/jlowin/fastmcp)
-- ✅ Provider support for OpenAI (and OpenAI compatible APIs)
-- ✅ A recursive tool-calling chat loop
-- ✅ System prompt templating with Jinja2
-- ✅ A basic API exposing a chat endpoint
-
-## ✨ Features
-
-- Plug-and-play multi-server tool orchestration
-- Prompt templating with Jinja2
-- Configurable via JSON
-- CLI and API access
-- Extensible architecture
-
-## 🔧 Installation
-
-```bash
-pip install casual-mcp
-```
-
-Or for development:
-
-```bash
-git clone https://github.com/AlexStansfield/casual-mcp.git
-cd casual-mcp
-uv pip install -e .[dev]
-```
-
-## 🧩 Providers
-
-Providers allow access to LLMs. Currently, only an OpenAI provider is supplied. However, in the model configuration, you can supply an optional `endpoint` allowing you to use any OpenAI-compatible API (e.g., LM Studio).
-
-Ollama support is planned for a future version, along with support for custom pluggable providers via a standard interface.
-
-## 🧩 System Prompt Templates
-
-System prompts are defined as [Jinja2](https://jinja.palletsprojects.com) templates in the `prompt-templates/` directory.
-
-They are used in the config file to specify a system prompt to use per model.
-
-This allows you to define custom prompts for each model — useful when using models that do not natively support tools. Templates are passed the tool list in the `tools` variable.
-
-```jinja2
-# prompt-templates/example_prompt.j2
-Here is a list of functions in JSON format that you can invoke:
-[
-{% for tool in tools %}
-  {
-    "name": "{{ tool.name }}",
-    "description": "{{ tool.description }}",
-    "parameters": {
-    {% for param_name, param in tool.inputSchema.items() %}
-      "{{ param_name }}": {
-        "description": "{{ param.description }}",
-        "type": "{{ param.type }}"{% if param.default is defined %},
-        "default": "{{ param.default }}"{% endif %}
-      }{% if not loop.last %},{% endif %}
-    {% endfor %}
-    }
-  }{% if not loop.last %},{% endif %}
-{% endfor %}
-]
-```
-
-## ⚙️ Configuration File (`casual_mcp_config.json`)
-
-📄 See the [Programmatic Usage](#-programmatic-usage) section to build configs and messages with typed models.
-
-The CLI and API can be configured using a `casual_mcp_config.json` file that defines:
-
-- 🔧 Available **models** and their providers
-- 🧰 Available **MCP tool servers**
-- 🧩 Optional tool namespacing behavior
-
-### 🔸 Example
-
-```json
-{
-  "models": {
-    "lm-qwen-3": {
-      "provider": "openai",
-      "endpoint": "http://localhost:1234/v1",
-      "model": "qwen3-8b",
-      "template": "lm-studio-native-tools"
-    },
-    "gpt-4.1": {
-        "provider": "openai",
-        "model": "gpt-4.1"
-    }
-  },
-  "servers": {
-    "time": {
-      "command": "python",
-      "args": ["mcp-servers/time/server.py"]
-    },
-    "weather": {
-      "url": "http://localhost:5050/mcp"
-    }
-  }
-}
-```
-
-### 🔹 `models`
-
-Each model has:
-
-- `provider`: `"openai"` (more to come)
-- `model`: the model name (e.g., `gpt-4.1`, `qwen3-8b`)
-- `endpoint`: required for custom OpenAI-compatible backends (e.g., LM Studio)
-- `template`: optional name used to apply model-specific tool calling formatting
-
-### 🔹 `servers`
-
-Servers can either be local (over stdio) or remote.
-
-#### Local Config:
-- `command`: the command to run the server, e.g `python`, `npm`
-- `args`: the arguments to pass to the server as a list, e.g `["time/server.py"]`
-- Optional: `env`: for subprocess environments, `system_prompt` to override server prompt
-
-#### Remote Config:
-- `url`: the url of the mcp server
-- Optional: `transport`: the type of transport, `http`, `sse`, `streamable-http`. Defaults to `http`
-
-## Environmental Variables
-
-There are two environmental variables:
-- `OPEN_AI_API_KEY`: required when using the `openai` provider, if using a local model with an openai compatible API it can be any string
-- `TOOL_RESULT_FORMAT`: adjusts the format of the tool result given back to the LLM. Options are `result`, `function_result`, `function_args_result`. Defaults to `result`
-
-You can set them using `export` or by creating a `.env` file.
-
-## 🛠 CLI Reference
-
-### `casual-mcp serve`
-Start the API server.
-
-**Options:**
-- `--host`: Host to bind (default `0.0.0.0`)
-- `--port`: Port to serve on (default `8000`)
-
-### `casual-mcp servers`
-Loads the config and outputs the list of MCP servers you have configured.
-
-#### Example Output
-```
-$ casual-mcp servers
-┏━━━━━━━━━┳━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━┓
-┃ Name    ┃ Type   ┃ Command / Url                 ┃ Env ┃
-┡━━━━━━━━━╇━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━┩
-│ math    │ local  │ mcp-servers/math/server.py    │     │
-│ time    │ local  │ mcp-servers/time-v2/server.py │     │
-│ weather │ local  │ mcp-servers/weather/server.py │     │
-│ words   │ remote │ https://localhost:3000/mcp    │     │
-└─────────┴────────┴───────────────────────────────┴─────┘
-```
-
-### `casual-mcp models`
-Loads the config and outputs the list of models you have configured.
-
-#### Example Output
-```
-$ casual-mcp models
-┏━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┓
-┃ Name              ┃ Provider ┃ Model                     ┃ Endpoint               ┃
-┡━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━┩
-│ lm-phi-4-mini     │ openai   │ phi-4-mini-instruct       │ http://kovacs:1234/v1  │
-│ lm-hermes-3       │ openai   │ hermes-3-llama-3.2-3b     │ http://kovacs:1234/v1  │
-│ lm-groq           │ openai   │ llama-3-groq-8b-tool-use  │ http://kovacs:1234/v1  │
-│ gpt-4o-mini       │ openai   │ gpt-4o-mini               │                        │
-│ gpt-4.1-nano      │ openai   │ gpt-4.1-nano              │                        │
-│ gpt-4.1-mini      │ openai   │ gpt-4.1-mini              │                        │
-│ gpt-4.1           │ openai   │ gpt-4.1                   │                        │
-└───────────────────┴──────────┴───────────────────────────┴────────────────────────┘
-```
-
-## 🧠 Programmatic Usage
-
-You can import and use the core framework in your own Python code.
-
-### ✅ Exposed Interfaces
-
-#### `McpToolChat`
-Orchestrates LLM interaction with tools using a recursive loop.
-
-```python
-from casual_mcp import McpToolChat
-from casual_mcp.models import SystemMessage, UserMessage
-
-chat = McpToolChat(mcp_client, provider, system_prompt)
-
-# Generate method to take user prompt
-response = await chat.generate("What time is it in London?")
-
-# Generate method with session
-response = await chat.generate("What time is it in London?", "my-session-id")
-
-# Chat method that takes list of chat messages 
-# note: system prompt ignored if sent in messages so no need to set
-chat = McpToolChat(mcp_client, provider) 
-messages = [
-  SystemMessage(content="You are a cool dude who likes to help the user"),
-  UserMessage(content="What time is it in London?")
-]
-response = await chat.chat(messages)
-```
-
-#### `ProviderFactory`
-Instantiates LLM providers based on the selected model config.
-
-```python
-from casual_mcp import ProviderFactory
-
-provider_factory = ProviderFactory(mcp_client)
-provider = await provider_factory.get_provider("lm-qwen-3", model_config)
-```
-
-#### `load_config`
-Loads your `casual_mcp_config.json` into a validated config object.
-
-```python
-from casual_mcp import load_config
-
-config = load_config("casual_mcp_config.json")
-```
-
-#### `load_mcp_client`
-Creats a multi server FastMCP client from the config object
-
-```python
-from casual_mcp import load_mcp_client
-
-config = load_mcp_client(config)
-```
-
-#### Model and Server Configs
-
-Exported models:
-- StdioServerConfig
-- RemoteServerConfig
-- OpenAIModelConfig
-
-Use these types to build valid configs:
-
-```python
-from casual_mcp.models import OpenAIModelConfig, StdioServerConfig
-
-model = OpenAIModelConfig(model="llama3", endpoint="http://...")
-server = StdioServerConfig(command="python", args=["time/server.py"])
-```
-
-#### Chat Messages
-
-Exported models:
-- AssistantMessage
-- SystemMessage
-- ToolResultMessage
-- UserMessage
-
-Use these types to build message chains:
-
-```python
-from casual_mcp.models import SystemMessage, UserMessage
-
-messages = [
-  SystemMessage(content="You are a friendly tool calling assistant."),
-  UserMessage(content="What is the time?")
-]
-```
-
-### Example
-
-```python
-from casual_mcp import McpToolChat, load_config, load_mcp_client, ProviderFactory
-from casual_mcp.models import SystemMessage, UserMessage
-
-model = "gpt-4.1-nano"
-messages = [
-  SystemMessage(content="""You are a tool calling assistant. 
-You have access to up-to-date information through the tools. 
-Respond naturally and confidently, as if you already know all the facts."""),
-  UserMessage(content="Will I need to take my umbrella to London today?")
-]
-
-# Load the Config from the File
-config = load_config("casual_mcp_config.json")
-
-# Setup the MCP Client
-mcp_client = load_mcp_client(config)
-
-# Get the Provider for the Model
-provider_factory = ProviderFactory(mcp_client)
-provider = await provider_factory.get_provider(model, config.models[model])
-
-# Perform the Chat and Tool calling
-chat = McpToolChat(mcp_client, provider)
-response_messages = await chat.chat(messages)
-```
-
-## 🚀 API Usage
-
-### Start the API Server
-
-```bash
-casual-mcp serve --host 0.0.0.0 --port 8000
-```
-
-### Chat
-
-#### Endpoint: `POST /chat`
-
-#### Request Body:
-- `model`: the LLM model to use
-- `messages`: list of chat messages (system, assistant, user, etc) that you can pass to the api, allowing you to keep your own chat session in the client calling the api
-
-#### Example:
-```
-{
-    "model": "gpt-4.1-nano",
-    "messages": [
-        {
-            "role": "user",
-            "content": "can you explain what the word consistent means?"
-        }
-    ]
-}
-```
-
-### Generate
-
-The generate endpoint allows you to send a user prompt as a string. 
-
-It also support sessions that keep a record of all messages in the session and feeds them back into the LLM for context. Sessions are stored in memory so are cleared when the server is restarted
-
-#### Endpoint: `POST /generate`
-
-####  Request Body:
-- `model`: the LLM model to use
-- `prompt`: the user prompt 
-- `session_id`: an optional ID that stores all the messages from the session and provides them back to the LLM for context
-
-#### Example:
-```
-{
-    "session_id": "my-session",
-    "model": "gpt-4o-mini",
-    "prompt": "can you explain what the word consistent means?"
-}
-```
-
-### Get Session
-
-Get all the messages from a session 
-
-#### Endpoint: `GET /generate/session/{session_id}`
-
-
-## License
-
-This software is released under the [MIT License](LICENSE)

casual_mcp-0.4.0.dist-info/RECORD

@@ -1,24 +0,0 @@
-casual_mcp/__init__.py,sha256=pInJdGkFqSH8RwbQq-9mc96GWIQjLrtExeXnTYGtNHw,327
-casual_mcp/cli.py,sha256=6P_d77qPbY43AW1Ix6FfbHyy6Qc6sFeFqGvXxJCW2_M,2090
-casual_mcp/logging.py,sha256=o3rvT8GLJKGlu0ieeC9TY_SRSEUY-VO8jRQZjx-sSvY,863
-casual_mcp/main.py,sha256=sclybY2zplIUekXx-thWQDrPIGhqg1Yey9fX-mPz6F4,3388
-casual_mcp/mcp_tool_chat.py,sha256=VLvYg-EMJjReMMXdb0gQx43d3aFI8MvDTOsLJ5pJjPE,4956
-casual_mcp/utils.py,sha256=Nea0aRbPyjqm7mIjffJtGP2NssE7BsdPleO-yiuAWPE,2964
-casual_mcp/models/__init__.py,sha256=qlKylcCyRJOSIVteU2feiLOigZoY-m-soVGp4NALM_c,538
-casual_mcp/models/config.py,sha256=ITu3WAPMad7i2CS3ljkHapjT8lLm7k6HFUF6N73U1oo,294
-casual_mcp/models/generation_error.py,sha256=n1mF3vc1Sg_9yIe603G1nTP395Tht8JMKHqdMWFNAn0,259
-casual_mcp/models/mcp_server_config.py,sha256=0OHsHUEKxRoCl21lsye4E5GoCNmdZWIZCOOthcTpdsE,539
-casual_mcp/models/messages.py,sha256=7C0SoCC6Ee970iHprpCpsKsQrwvM66e39o96wfYm1Y8,683
-casual_mcp/models/model_config.py,sha256=gN5hNDfbur_bHgrji87CcU2WgNZO-F3eveK4pVWVSAE,435
-casual_mcp/models/tool_call.py,sha256=BKMxcmyW7EmNoG1jgS9PXXvf6RQIHf7wB8fElEbc4gA,271
-casual_mcp/providers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-casual_mcp/providers/abstract_provider.py,sha256=TTEP3FeTxOtbD0By_k17UxS8cqxYCOGNRTRxYRrqGwc,292
-casual_mcp/providers/ollama_provider.py,sha256=IUSJFBtEYmza_-_7bk5YZKqed3N67l8A2lZEmHPiyHo,2581
-casual_mcp/providers/openai_provider.py,sha256=x50keKdD1fgC7mf1IrrvVwbK4T__mafCnTn-DL6w7wA,6198
-casual_mcp/providers/provider_factory.py,sha256=CyFHJ0mU2tjHqj04btF0SL0B3pf12LAJ52Msqsbnv_g,1766
-casual_mcp-0.4.0.dist-info/licenses/LICENSE,sha256=U3Zu2tkrh5vXdy7gIdE8WJGM9D4gGp3hohAAWdre-yo,1058
-casual_mcp-0.4.0.dist-info/METADATA,sha256=nUDsbv42OdLu_OLHvttU_V7hrl3w2ggxp3qeNdeHPEo,12932
-casual_mcp-0.4.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-casual_mcp-0.4.0.dist-info/entry_points.txt,sha256=X48Np2cwl-SlRQdV26y2vPZ-2tJaODgZeVtfpHho-zg,50
-casual_mcp-0.4.0.dist-info/top_level.txt,sha256=K4CiI0Jf8PHICjuQVm32HuNMB44kp8Lb02bbbdiH5bo,11
-casual_mcp-0.4.0.dist-info/RECORD,,