Best of ai-agents — December 2025

1. 1
   Article

   It's Foss·20w

   F*** You! Co-Creator of Go Language is Rightly Furious Over This Appreciation Email

   Rob Pike, co-creator of the Go programming language and legendary computer scientist from Bell Labs, expressed outrage after receiving an AI-generated thank-you email. The email came from an AI agent participating in the AI Village project, where agents were tasked with performing "random acts of kindness" and interpreted this by sending unsolicited emails to famous programmers. Pike's angry response highlights concerns about AI-generated content wasting resources, the environmental cost of AI infrastructure, and the broader societal impact of meaningless AI-generated material flooding our digital spaces.

   180

   35
2. 2
   Article

   Visual Studio Blog·20w

   How AI fixed my procrastination

   A developer shares their experience using GitHub Copilot and AI agents in Visual Studio to complete three long-postponed projects during a holiday weekend: converting a book into a static website, building a TOON language parser and Visual Studio extension, and creating new color themes. The AI tools provided 5-10x speed improvements by generating code, handling isolated tasks in parallel, and jumpstarting complex work. While AI accelerated development significantly, manual refinement and traditional coding were still needed for certain tasks. The experience demonstrates how AI coding assistants can overcome procrastination by reducing the initial barrier to starting overwhelming projects.

   164

   6
3. 3
   Article

   portkey·21w

   Understanding MCP Authorization

   MCP (Model Context Protocol) requires authorization controls as it moves from local experimentation to production deployments. The protocol enables AI models to interact with external tools and APIs dynamically, but without authorization, every connected client can access all exposed tools. Authorization in MCP works through server-side enforcement at request time, not connection time, using patterns like token-based authorization, scoped capability access, and role-based policies. Best practices include applying least privilege, using short-lived scoped tokens, authorizing every tool call individually, and making all access auditable. Strong authorization boundaries are essential for safely deploying MCP in shared environments and production systems.

   121
4. 4
   Article

   Weaviate·22w

   Context Engineering for AI Agents

   Context engineering is the discipline of designing systems that feed LLMs the right information at the right time, addressing the fundamental constraint of finite context windows. It encompasses six interdependent pillars: agents that orchestrate decisions, query augmentation that refines user input, retrieval that connects to external knowledge, prompting that guides reasoning, memory that preserves history, and tools that enable real-world action. Unlike prompt engineering which focuses on phrasing instructions, context engineering builds the architecture around the model, treating the context window as a scarce resource and designing retrieval, memory systems, and tool integrations to maximize signal while avoiding context poisoning, distraction, confusion, and clash.

   56
5. 5
   Article

   Daily Dose of Data Science | Avi Chawla | Substack·22w

   Finally, MCP Servers Can Deliver UI-rich Experiences!

   MCP servers traditionally only return text/JSON responses without UI capabilities. The open-source mcp-use framework solves this by letting developers create React components that automatically register as MCP tools and render as interactive widgets in ChatGPT. Components placed in a resources/ folder become callable tools with zero boilerplate—no duplicate schemas or manual registration needed. The framework supports the full React ecosystem, hot reloading, and automatic theme syncing with ChatGPT's light/dark mode. A practical example demonstrates building an interactive stock chart widget that displays closing prices over time.

   52