Tag
OpenAI announces that users can now build and deploy MCP servers directly through ChatGPT via ChatGPT Sites, with access controls and sharing options for the deployed servers and plugins.
The article presents an architectural mediation approach using the Model Context Protocol to enable controlled interactions between LLM agents and data spaces, ensuring compliance with governance and interoperability standards.
The article argues that MCP, originally useful for connecting LLMs to external services, has become outdated due to advancements in LLMs that can now handle API calls directly, rendering many MCP servers unnecessary.
This article describes Stars!VM, a custom virtual machine that emulates an 80286 processor and bridges Win16 to Win32 to play the 1995 4X game Stars! on modern systems with features like 4k scaling.
This paper introduces a closed-world resolution method to combat tool hallucination in LLM agents, offering a taxonomy and benchmark for measuring and addressing fabricated tool calls.
Kody v2026.09.09 is released, featuring a new documentation site with a sticky grouped sidebar, sectioned reading order, and dedicated concept docs for AI agent primitives.
Observation that Blender's integration with the Model Context Protocol is disrupting the 3D design economy.
A Hacker News user asks about production usage of Model Context Protocol (MCP), with comments discussing its advantages and examples from real-world applications like UI updates and open data initiatives.
The paper presents an agentic AI framework using large language models and the Model Context Protocol to automate atomic force microscope operation for nanoscale characterization, achieving expert-level performance with safe execution.
This paper presents a language-model-based workflow for generating, validating, and correcting ABB RAPID robot programs using retrieval-augmented generation and Model Context Protocol, validated through simulation in ABB RobotStudio.
The Model Context Protocol (MCP) has published an updated roadmap outlining five priority areas for development, including agentic messaging primitives, HTTP-native transport unification, and agent identity security. The roadmap sets the direction for protocol work over the coming months, developed by core maintainers and the community.
Agentao introduces a governed local-first runtime for tool-using LLM agents, separating model-generated actions from host-authorized execution to improve safety and governance.
Qencode MCP lets AI agents transcode and process video, integrating video transcoding capabilities into agent workflows.
The article discusses the deprecation of Sampling in the MCP spec as of the 2026-07-28 changelog, shifting model-call costs from clients to servers, and advises how to check if a server relies on Sampling via code or logs.
Raises concerns that MCP servers may introduce new architectural dependencies, questioning whether agents tied to specific server auth and implementations are truly portable.
This paper introduces 'canary tools' as diagnostic probes to identify specific tool-selection reasoning failures in LLM agents, proposing a six-type taxonomy and evaluating eight models across 8,640 task runs to show capability-tier disparities and robustness.
An opinion piece arguing that MCP servers often fail because teams build them without confirming real user demand, and recommends validating with actual struggling users and early usage instrumentation before committing.
Simon Willison discusses the stateless MCP (MCP 2.0) specification, which simplifies protocol implementations, and introduces his new tools mcp-explorer and datasette-mcp built to leverage it.
Simon Willison announces the initial alpha release of llm-mcp-client, a Python client for the Model Context Protocol.
Dan Kornas highlights a curated directory of Model Context Protocol (MCP) servers, grouping implementations by job and linking to supported hosts like Claude Desktop, Cursor, VS Code, and Zed. The open-source directory includes security guidance and official markers.