Merge pull request #9 from parente/parente/mcp

Model Context Protocol (MCP) Brief

Author: parente

Makefile

@@ -16,13 +16,14 @@ clean: ## Make a clean workspace
 	@git clean -f .
 
 check: ## Make a ruff check of code lint
-	@ruff check generate.py
+	@uv run ruff check generate.py
 
 env: ## Make the current python environment install all prereqs
-	@pip install -r requirements.txt -r requirements-dev.txt
+	@uv venv
+	@uv pip install -r requirements.txt -r requirements-dev.txt
 
 build: ## Make a local copy of the blog
-	python generate.py
+	@uv run python generate.py
 
 server: ## Make a local web server point to the latest local build
 	@open http://localhost:8000/_output && python -m http.server

pages/20250329-mcp-brief/index.md

@@ -0,0 +1,49 @@
+---
+date: 2025-03-29
+title: Model Context Protocol (MCP) Brief
+excerpt:
+    I've been hearing about Model Context Protocol (MCP) lately. I wrote this brief to connect MCP
+    to my knowledge about LLM function calling and my experience implementing tools for Anthropic Claude integrations at work.
+---
+
+I've implemented a handful of
+[tools](https://docs.anthropic.com/en/docs/build-with-claude/tool-use/overview) for Anthropic Claude
+integrations at work: fetching public web page content, invoking Slack APIs, querying Amazon Bedrock
+Knowledge Bases. I've been hearing about [Model Context Protocol
+(MCP)](https://modelcontextprotocol.io) lately, and haven't had time to grok it's connection to LLM
+function calling and my experience.
+
+With a good cup of ☕️ coffee and some time to read and think, I've arrived at the following
+understanding in brief.
+
+---
+
+-   Many LLMs support _function calling_ also known as _tool use_ (e.g., [OpenAI](https://platform.openai.com/docs/guides/function-calling), [Anthropic](https://docs.anthropic.com/en/docs/build-with-claude/tool-use/overview), [Llama](https://www.llama.com/docs/model-cards-and-prompt-formats/llama3_2/#code-interpreter)).
+-   There's no standard format for function call / tool use instructions across models vendors. Tool
+    request and response formats vary.
+-   MCP "... provides a standardized framework for managing the execution of function call instructions,
+    including tool discovery, invocation, and response handling"[^chan2024dec]
+-   [MCP servers](https://modelcontextprotocol.io/examples) host tools discovery and execution.
+-   [MCP clients](https://modelcontextprotocol.io/clients) bridge user interaction with LLMs, and
+    LLM interactions with tools on MCP servers. For example:
+    -   User: "What's the weather?"
+    -   MCP client: List tools available for LLM use
+    -   LLM: Weather tool use request in LLM format
+    -   MCP client: JSON-RPC call to weather function
+    -   MCP server: Response to weather function call
+    -   MCP client: Weather tool use response in LLM format
+    -   LLM: "The weather is bright and sunny"
+-   MCP enables an ecosystem of adapters for existing services (e.g., [https://mcp.so/](https://mcp.so)).
+-   MCP uses [JSON-RPC 2.0](https://www.jsonrpc.org/specification) on the wire with support for local (stdio) and remote ([HTTP Server-Sent Events](https://en.wikipedia.org/wiki/Server-sent_events)) transports.[^mcp2025]
+-   It supports concepts beyond tools (e.g., [resources](https://modelcontextprotocol.io/docs/concepts/resources), [prompts](https://modelcontextprotocol.io/docs/concepts/prompts)).
+-   It fits into existing model broker architectures (e.g., [Amazon Bedrock](https://docs.aws.amazon.com/bedrock/latest/userguide/)[^battista2025mar], [Ollama](https://ollama.com/)[^gupta2025mar]).
+-   There are security, cost, development, hosting, pricing (and likely other) implications of the
+    widespread adoption of MCP.[^li2025mar]
+
+# References
+
+[^chan2024dec]: Chan, P. (2024 Dec 13). [LLM Function-Calling vs. Model Context Protocol (MCP)](https://medium.com/@patc888/function-calling-vs-mcp-model-context-protocol-fc01e9c41eb4). _Patrick Chan on Medium_.
+[^mcp2025]: [Transports](https://modelcontextprotocol.io/docs/concepts/transports). (2025 March 29). In _Model Context Protocol_.
+[^battista2025mar]: Battista, G. (2025 Mar 19). [Model Context Protocol (MCP) and Amazon Bedrock](https://community.aws/content/2uFvyCPQt7KcMxD9ldsJyjZM1Wp/model-context-protocol-mcp-and-amazon-bedrock). _AWS Community._
+[^gupta2025mar]: Gupta, M. (2025 Mar 29). [Model Context Protocol (MCP) using Ollama](https://medium.com/data-science-in-your-pocket/model-context-protocol-mcp-using-ollama-e719b2d9fd7a). _Data Science in Your Pocket_.
+[^li2025mar]: Li, Y. (2025 March 20) [A Deep Dive Into MCP and the Future of AI Tooling](https://a16z.com/a-deep-dive-into-mcp-and-the-future-of-ai-tooling/). _Andreesen Horowitz_.