Model Context Protocol

Prompt Engineering Retrieval-Augmented Generation
Anthropic released a Data Standard and an Open Specification
Fastify
https://youtu.be/f4mI3d-nTrI?is=kkIpNiMtv8cL88Og
https://youtu.be/RhTiAOGwbYE?si=utXq40hwAmpS8sCw
https://youtu.be/sF799nFJONk?si=bzKPSbS1k_JjbVXo
by Sean Kochel
https://www.youtube.com/watch?v=ixc9t7wTe6U
https://youtu.be/oAoigBWLZgE?si=8AF2ooyl_EMhe2sM
https://youtu.be/g08kmknV5Sg?si=1VzgKs5TjxMeoWnY
https://youtube.com/shorts/pFM16nVrnD4?si=zw_LOqOKCns_zsxY
https://youtu.be/japoGcdbZGw?si=RE2kLXYfNygeuvOd
https://youtu.be/m46tZX6vceI?si=Xhuz94XNYIa1y3Xp
https://youtu.be/TcNd6rTOpJA?si=R-DAf8H-dhfbPiVc
https://youtu.be/7j_NE6Pjv-E?si=CzSqldFv18gN37Q0
https://youtu.be/EEE-l41_VQ0?si=JiQE4mG3v0OrVhLE
https://youtu.be/sVC4DL2secQ?si=u9UQE9UrxL0nvY-G
https://youtu.be/nNLshWCoe0o?si=LXuc5U8142jU2y67
https://youtu.be/qYChSSP8TTA?si=Leah7CGH_oGHbfK0
https://youtu.be/5P4IJS9q_yg?si=2kp7Hm7cT7XrgkZ7
https://youtu.be/GITdDHglRfg?si=jKEoyh6oVVGGX-x5
https://youtu.be/kQmXtrmQ5Zg?si=G49rIFWBuBbURE6T
https://youtu.be/TQsP_PlCY1I?si=1LyMWwn5gB00xUf1
https://youtu.be/yOKwK-iIg3M?si=HJzKK3y8mQTTB5F
_
https://youtu.be/MC2BwMGFRx4?si=_9zpDQOAv-XB0_h0
https://youtu.be/sMqlObpNz64?si=bPm5L8QQbY8YXYN
_
https://youtu.be/3Jsh4brTjE0?si=dLjkIt5qDoKCgBmy
https://youtu.be/_rFissIE6CA?si=8bNO4zfc57lcJSjn
https://youtu.be/m46tZX6vceI?si=7CpwyesS6mTxEXfk
https://youtu.be/oAoigBWLZgE?si=8AF2ooyl_EMhe2sM
https://youtube.com/shorts/pFM16nVrnD4?si=zw_LOqOKCns_zsxY
https://youtu.be/japoGcdbZGw?si=RE2kLXYfNygeuvOd
https://youtu.be/m46tZX6vceI?si=Xhuz94XNYIa1y3Xp
https://youtu.be/GITdDHglRfg?si=jKEoyh6oVVGGX-x5
https://youtu.be/kQmXtrmQ5Zg?si=G49rIFWBuBbURE6T
https://youtu.be/TQsP_PlCY1I?si=1LyMWwn5gB00xUf1
https://youtu.be/yOKwK-iIg3M?si=HJzKK3y8mQTTB5F
_
https://youtu.be/MC2BwMGFRx4?si=_9zpDQOAv-XB0_h0
https://youtu.be/sMqlObpNz64?si=bPm5L8QQbY8YXYN
_
https://youtu.be/3Jsh4brTjE0?si=dLjkIt5qDoKCgBmy
https://youtu.be/_rFissIE6CA?si=8bNO4zfc57lcJSjn
https://youtu.be/m46tZX6vceI?si=7CpwyesS6mTxEXfk
https://youtu.be/eur8dUO9mvE?si=SdJgtKl4m4xrbIgk
ℹ️ (2025-08-17T19:09.457Z)
Perplexity Query
(2025-08-17T19:09
.457Z)
Question:
When did Model Context Protocol come out? Anthropic released the spec but who created it? There seem to be a million "MCP Servers" but how do we know if they are useful? How would developers best use public MCP Servers? What would it take to create one?

Response from sonar-pro :

The Model Context Protocol (MCP) was introduced by Anthropic in November 2024 as an open standard to unify and simplify integration between AI models and external systems. [nxre0z] [wzp5sx] The protocol was created by Anthropic engineers David Soria Parra and Justin Spahr-Summers . [frxou3]
Anthropic released a Data Standard

Overview and Origins

  • Release Date: November 2024. [nxre0z] [wzp5sx]
  • Primary Creators: David Soria Parra and Justin Spahr-Summers of Anthropic. [frxou3]
  • Purpose: MCP standardizes how AI systems share and manage contextual data, replacing fragmented custom integrations with a unified, versioned, and extensible approach. [5atjh6] [nxre0z] [wzp5sx]

MCP Servers: Usefulness and Proliferation

There are indeed many MCP Servers , often created by different vendors, open-source contributors, or teams to expose specific integrations (for example, GitHub access, database operations, code search, or file management ). [4rpboa] [wzp5sx]

Assessing Usefulness

To determine if an MCP server is useful:
  • Documentation and Specification Compliance: Check if the server fully implements the MCP spec, supports required transport methods (STDIO and HTTP/SSE), and exposes clear, well-documented capabilities. [4rpboa]
  • Community Adoption and Maintenance: Frequent updates, number of users, and support in major host applications (e.g., Claude Desktop, Zed) signal reliability. [wzp5sx]
  • Security and Permissions: Look for support for robust authentication and permission controls.
  • Ecosystem Integration: Plug-and-play compatibility with popular MCP clients and seamless performance in real workflows.
IMAGE 2: Practical example or use case visualization (A mockup here would show a developer connecting their LLM-based IDE to a "GitHub MCP server" that provides repository access, issue lookup, and pull request integration.)

Best Practices for Developers Using Public MCP Servers

  • Discover and Select: Explore registry listings or curated directories for MCP servers; prioritize popular, well-maintained options.
  • Audit the Capabilities: Understand which functions each server exposes and ensure they match your project needs.
  • Security First: Use servers that support authentication, sandboxing, and audit logging, especially for sensitive integrations.
  • Version Management: MCP supports versioned context objects—developers should ensure compatibility between their application and the server.
  • Local vs. Remote: Use STDIO for local, high-trust use cases; HTTP+SSE for remote, cloud-native setups. [4rpboa]
  • Community Engagement: Participate in the MCP ecosystem (e.g., submit bug reports, contribute integrations, attend workshops).

What It Takes to Create an MCP Server

To build an MCP server:
  • Implement MCP Spec: Follow the latest Anthropic MCP specification, including support for:
    • JSON-RPC 2.0 as the message protocol. [4rpboa]
    • Transport Layers: Support for STDIO (local) and HTTP + SSE (remote streaming). [4rpboa]
  • Define Exposed Capabilities: Decide on the functions/data your server will expose—e.g., access to a data source, code repository, or external API. [5atjh6] [4rpboa]
  • Write Server Logic: Build in the desired language (MCP is language-agnostic; common choices include Python, Node.js, Go).
  • Maintain Documentation: Clearly document methods, authentication flows, and version compatibility.
  • Test and Validate: Use MCP clients to test interoperability; ensure robust error handling and extensibility.
Summary Table
Aspect Details
Protocol Name Model Context Protocol (MCP)
Released November 2024
Creators David Soria Parra, Justin Spahr-Summers (Anthropic)
Protocol Purpose Standardizes model/context integration, reduces friction
Server Usefulness Spec compliance, adoption, security, documentation
Developer Best Use Audit, secure, version, local/remote settings
Build Requirements MCP spec, JSON-RPC 2.0, transport, documented functions
MCP’s core value is its open, standardized, and extensible architecture , making it straightforward for developers to integrate diverse AI tools and data sources into robust applications. [5atjh6] [nxre0z] [4rpboa] [wzp5sx]

Citations

[5atjh6] 2025, Apr 17. What is the Model Context Protocol (MCP), and why was it ... . Published: 2025-04-09 | Updated: 2025-04-17

[nxre0z] 2025, Aug 17. Model Context Protocol . Published: 2025-04-14 | Updated: 2025-08-17

[frxou3] 2025, Jul 05. MCP Protocol: a new AI dev tools building block . Published: 2025-04-08 | Updated: 2025-07-05

[4rpboa] 2025, Jul 19. What Is the Model Context Protocol (MCP) and How It Works . Published: 2025-04-07 | Updated: 2025-07-19

[wzp5sx] 2025, Jul 08. A beginners Guide on Model Context Protocol (MCP) . Published: 2025-04-03 | Updated: 2025-07-08