# Agentailor > Agentailor is a technical blog by Ali Ibrahim, part of agentailor.com — the hub for developers building AI agents. The blog publishes in-depth, code-first articles on building AI agents across all four paths: building it yourself (with frameworks like LangGraph or the OpenAI Agents SDK), building it with a coding agent (agentic engineering with tools like Claude Code), deploying an open-source agent (such as Goose), and using a managed service. It covers the core agent primitives — models, tools, memory, prompting, and the decision loop — along with the Model Context Protocol (MCP), context engineering, agentic engineering, evaluation criteria for choosing an approach, and production considerations. The target audience is software engineers and AI engineers, from those new to agents to experienced practitioners with specific architectural needs. All articles include working code examples. Note: All article links point to .md files containing clean Markdown. Use these instead of the HTML pages to save tokens. ## Series Ongoing case studies where each entry builds on the last, rather than standalone tutorials. Start at the hub for the full arc. ### [Cameron](https://blog.agentailor.com/cameron) A local-first personal finance agent you own and run yourself — an AI agent you can trust with real money because it never acts without your approval. Built in public, one tagged release per article: each entry adds a capability to the same agent instead of starting over, so you learn agent engineering by watching one real system grow. Hub: https://blog.agentailor.com/cameron — the canonical page for this series (roadmap, design invariants, and every entry). Entries, in reading order: - v1 — [Meet Cameron: a personal finance agent you own and run](https://blog.agentailor.com/posts/meet-cameron.md) (devlog): Cameron is a local-first personal finance agent you own and run yourself. v1 shows how an LLM loop, tools, and an approval gate compose into an agent you can trust with real money. - v2 — [Testing Tools for AI Agents: catching the bugs your agent can't tell you about](https://blog.agentailor.com/posts/testing-tools-for-ai-agents.md) (devlog): You designed the tool well. Nothing verifies it still does what its description promises. An audit of eight tools found four contract bugs, including one that made a finance agent report $2,812 for a $2,752 bill, and the cheap test layer that catches them. - v2 — [How to Write Your First AI Agent Evals](https://blog.agentailor.com/posts/first-eval-cases.md) (devlog): Where the first cases come from, which ones earn a place in the suite, why every case that pushes a behavior needs one that bounds it, how to grade without an LLM judge, and what the first honest run surfaced. - v2.1 — [When Your Eval Script Answers a Question Nobody Asked](https://blog.agentailor.com/posts/simulated-users-for-agent-evals.md) (devlog): A scripted multi-turn eval replays fixed user turns regardless of what the agent said. The moment the agent asks a question the case author did not anticipate, the script answers something else and the case fails for the wrong reason - rewarding agents that guess and punishing agents that ask. - v3 — [How to Add Agent Skills Support to a Custom Agent Harness](https://blog.agentailor.com/posts/how-to-add-skills-to-a-custom-agent.md): Add Agent Skills support to your custom AI agent: SKILL.md discovery, progressive disclosure, tool loading, and evals, with TypeScript and Python examples. ## Reports Agentailor Reports answer one question each with data we collected: the method is fixed before the results, and every number recomputes from a public repository. Index: https://blog.agentailor.com/reports - [AR-001: On a real agent's eval rubrics, can Jev replace the LLM judges?](https://blog.agentailor.com/reports/ar-001-can-jev-replace-llm-judges.md): Not as a straight swap. As a confident first pass in front of Sonnet, it matches Sonnet's accuracy at about a sixth of the cost. ## Blog Posts - [Before You Add WebMCP: the Security Problem Nobody's Selling You](https://blog.agentailor.com/posts/before-you-add-webmcp.md): WebMCP lets a web page hand tools to AI agents, and most coverage sells it as your website becoming an API. The spec, WebKit, and even ChatGPT say something the pitch leaves out: an agent cannot verify what a page tool does, and it runs with your users logged-in session. I shipped both sides of WebMCP on Agentailor, red-teamed them, and deleted one. Here is what that taught me. - [How to Add Agent Skills Support to a Custom Agent Harness](https://blog.agentailor.com/posts/how-to-add-skills-to-a-custom-agent.md): Add Agent Skills support to your custom AI agent: SKILL.md discovery, progressive disclosure, tool loading, and evals, with TypeScript and Python examples. - [Building an Agent Eval Harness: The Complete Guide](https://blog.agentailor.com/posts/agent-eval-harness.md): An eval harness is four components: the case, the runner, the capture, and the graders. A guide to what each one has to do for an AI agent, built around two real harnesses, so you can decide whether to build one, adopt a framework, or combine both. - [Your System Prompt Has a Shelf Life: Maintaining Prompts as Models Improve](https://blog.agentailor.com/posts/system-prompt-shelf-life.md): Anthropic deleted over 80% of Claude Code system prompt for Claude 5 with no measurable loss on their coding evals. The instructions did not become wrong. The model outgrew needing them. Here is the maintenance pass that finds which of yours have expired, and why you need evals before you delete a single line. - [Agent Plugins Explained: Skills and MCP in One Folder](https://blog.agentailor.com/posts/agent-plugins-explained.md): Agent Plugins is an open standard for packaging Agent Skills and MCP servers into one portable folder any client can load. What plugin.json and mcp.json actually define, which clients support it, and whether to package one yet. - [How to Write AI Agent Evals That Prove You Wrong](https://blog.agentailor.com/posts/how-to-write-ai-agent-evals.md): Write AI agent evals that catch what you missed: source cases from production traces, spot tests that pass for the wrong reason, and know when to overrule an LLM judge. - [Securing an MCP v2 Server with OAuth and Keycloak](https://blog.agentailor.com/posts/mcp-v2-oauth-keycloak.md): What changes for OAuth when you move an MCP server to v2. The MCP TypeScript v2 SDK moves the auth helpers into @modelcontextprotocol/express and surfaces auth info via ctx.http.authInfo — your Keycloak setup stays exactly the same. - [Build an MCP v2 Server in TypeScript with the New MCP SDK](https://blog.agentailor.com/posts/build-an-mcp-v2-server-typescript.md): Build a stateless MCP v2 server in TypeScript. A hands-on guide to the MCP TypeScript v2 SDK — the new @modelcontextprotocol/server packages, createMcpHandler, Standard Schema tools, and what replaces session handling. - [MCP v2: What's Changing, What's Deprecated, and Why](https://blog.agentailor.com/posts/mcp-v2-whats-changing-and-why.md): MCP v2 breaking changes explained: the protocol goes stateless and deprecates sampling, roots, and logging. What changed across every MCP SDK, why, and how to plan your migration. - [Is LangChain Worth It in 2026? A Practitioner's Guide From Building to Production](https://blog.agentailor.com/posts/is-langchain-worth-it-2026.md): Is LangChain worth it in 2026? A practitioner's guide to the v1 stack: create_agent, middleware, deploying, observing, and MCP, with an honest verdict. - [Top 7 AI Agent Evaluation Frameworks (2026)](https://blog.agentailor.com/posts/top-ai-agent-eval-frameworks-2026.md): Discover the top open-source AI agent evaluation frameworks in 2026. From DeepEval and promptfoo to agentevals and Strands Evals, see which eval tools to use for trajectory checking, CI gating, red-teaming, and scoring agents in production. - [Top Agent Skills for Agentic Engineering (and Spec-Driven Development)](https://blog.agentailor.com/posts/top-agent-skills-for-agentic-engineering-2026.md): Frontier coding agents are capable but non-deterministic. These are the agent skills for agentic engineering and spec-driven development that keep their work durable across sessions. - [In 2026, There Are 4 Ways to Build an AI Agent. Here's How to Choose.](https://blog.agentailor.com/posts/four-ways-to-build-an-ai-agent-2026.md): Building an AI agent in 2026 isn't one decision — it's four different paths. Here's how to tell which one fits what you're actually trying to do. - [You Don't Need to Build Your AI Agent — Just Use Goose](https://blog.agentailor.com/posts/goose-open-source-agent-quickstart.md): Most agent content teaches you to build. Goose is a battle-tested open-source AI agent you can install and run today, with 70+ extensions and the full codebase if you ever need to go deeper. - [Top 5 AI Agent Protocols to Know in 2026](https://blog.agentailor.com/posts/top-ai-agent-protocols-2026.md): Frameworks are the engines of AI agents, but protocols are the roads they travel. Here are the 5 most important AI agent protocols shaping the ecosystem in 2026. - [Top AI Agent Standards to Know in 2026](https://blog.agentailor.com/posts/top-ai-agent-standards-2026.md): Protocols tell agents how to connect. Standards tell them what to know. Here are the open formats giving agents consistent context about projects, capabilities, and design systems in 2026. - [Observability for AI Agents: Why Tracing Matters and How to Do It with Langfuse](https://blog.agentailor.com/posts/observability-tracing-ai-agents-langfuse.md): AI agents fail in ways that infrastructure traces can't explain. Learn why production agents need semantic observability, and how to add it to a LangGraph + Next.js app with Langfuse. - [How I Made My Blog Native to AI Agents (And Launched One)](https://blog.agentailor.com/posts/how-i-made-my-blog-native-to-ai-agents.md): The web was built for browsers. I spent the last few months making Agentailor native to AI agents too — here is what I built, why I built it, and what you can steal for your own project. - [AI Agent Roadmap: Everything You Need to Build Agents (In the Right Order)](https://blog.agentailor.com/posts/agent-development-roadmap.md): A step-by-step roadmap for developers building AI agents — covering LangGraph, MCP, tools, memory, evals, and production deployment in Python and TypeScript. - [The MCP TypeScript SDK: A Complete Guide to Tools, Resources, Prompts, and Beyond](https://blog.agentailor.com/posts/mcp-typescript-sdk-complete-guide.md): Go beyond the basics of MCP server development. A comprehensive guide to the official TypeScript SDK — architecture, tools, resources, prompts, sampling, completions, and more. - [Top 5 Agent Skills Every Agent Builder Should Install](https://blog.agentailor.com/posts/top-agent-skills-for-agent-builders-2026.md): The 5 most impactful agent skills for developers building AI agents with coding IDEs. From prompt engineering to agent evaluation, each skill includes what it does, why it matters, and the install command. - [Deploy Your MCP Server to Google Cloud Run (For Free)](https://blog.agentailor.com/posts/deploy-mcp-server-cloud-run.md): Take your MCP server from localhost to the cloud in minutes. Deploy to Google Cloud Run using the free tier, no local Docker required. - [Lessons from OpenClaw's Architecture for Agent Builders](https://blog.agentailor.com/posts/openclaw-architecture-lessons-for-agent-builders.md): OpenClaw went from a weekend script to 200k+ GitHub stars. This architectural deep dive breaks down the gateway pattern, Lane Queue system, skills-as-markdown extensibility, and security lessons every agent builder needs. - [How to Build and Deploy an Agent Skill from Scratch](https://blog.agentailor.com/posts/how-to-build-and-deploy-agent-skill-from-scratch.md): AI agents are only as useful as the skills they have. In this guide, we build a complete Agent Skill from scratch—a data visualization skill that generates financial reports with Chart.js and deploy it to a personal finance assistant called Cameron AI. - [Getting Started with FastMCP in TypeScript](https://blog.agentailor.com/posts/getting-started-with-fastmcp.md): Build MCP servers faster with FastMCP — the TypeScript framework inspired by Python's most popular MCP library. A complete guide to scaffolding and building your first FastMCP server. - [Implementing OAuth for MCP Clients: A Next.js + LangGraph.js Guide](https://blog.agentailor.com/posts/mcp-client-oauth-nextjs-langgraph.md): Learn how to implement OAuth authentication in your MCP client. A practical guide for connecting AI agents to secured MCP servers using Next.js and the MCP SDK. - [Create Your First MCP Server in 5 Minutes with create-mcp-server](https://blog.agentailor.com/posts/create-your-first-mcp-server-in-5-minutes.md): Build your first production-ready MCP server in under 5 minutes. A complete beginner guide to scaffolding a Fetch MCP server with TypeScript using create-mcp-server. - [Securing MCP Servers: A Practical Guide with Keycloak (using create-mcp-server)](https://blog.agentailor.com/posts/oauth-for-mcp-servers-practical-guide-keycloak.md): Learn how to secure your MCP servers with OAuth authentication using Keycloak. A hands-on guide using create-mcp-server to scaffold production-ready, authenticated MCP servers. - [Top 10 Most Starred AI Agent Frameworks on GitHub (2026)](https://blog.agentailor.com/posts/top-ai-agent-frameworks-github-2026.md): Discover the top 10 most popular AI agent frameworks ranked by GitHub stars. From LangChain to Mastra, see which tools developers trust most for building AI agents. - [Top 5 TypeScript AI Agent Frameworks You Should Know in 2026](https://blog.agentailor.com/posts/top-typescript-ai-agent-frameworks-2026.md): Discover the top TypeScript AI agent frameworks ranked by GitHub stars and npm downloads. From Vercel AI SDK to LangGraph.js. - [Engineering Context for Local and Cloud AI: Personas, Content Intelligence, and Zero-Prompt UX](https://blog.agentailor.com/posts/context-engineering-hybrid-ai-personas.md): How DocuMentor AI delivers personalized results without asking users what they want. A deep dive into persona-driven design, strategic content decomposition, and zero-prompt UX for hybrid AI systems. - [Engineering a Hybrid AI System with Chrome’s Built‑in AI and the Cloud](https://blog.agentailor.com/posts/chrome-builtin-ai-cloud-hybrid-system.md): Learn how to build production-ready AI features that adapt seamlessly between Chrome's local Gemini Nano and cloud models. Based on building the DocuMentor AI Chrome extension, this deep dive covers provider abstraction, Chrome AI resource management, execution patterns for local vs cloud, and graceful fallback systems. - [Writing Effective Tools for AI Agents: Production Lessons from Anthropic](https://blog.agentailor.com/posts/writing-tools-for-ai-agents.md): Anthropic's Applied AI team reveals how to design tools that agents can actually use effectively. Learn the five production-tested principles for building agent tools, with a complete implementation of Cameron AI's expense tracking system. - [The Art of Agent Prompting: Anthropic’s Playbook for Reliable AI Agents](https://blog.agentailor.com/posts/the-art-of-agent-prompting.md): Learn Anthropic’s production-tested playbook for prompting AI agents like Claude Code. We distill their principles, heuristics, and eval strategies into a practical guide you can use to build reliable, tool-using agents for the real world. - [Building a Fullstack AI Agent with LangGraph.js and Next.js: MCP Integration & Human-in-the-Loop](https://blog.agentailor.com/posts/fullstack-ai-agent-app-with-langgraphjs-and-nextjs.md): Learn how to build a production-ready fullstack AI agent with Next.js and LangGraph.js. This guide covers Model Context Protocol integration for dynamic tool loading, human-in-the-loop approval workflows, and real-time streaming with React Query. - [How to Publish Your MCP Server to the Official Registry: A Complete Guide](https://blog.agentailor.com/posts/publishing-mcp-server-to-official-registry.md): Learn how to publish your MCP server to the official Model Context Protocol registry with step-by-step instructions and real-world examples. - [Don’t Let Your AI Agent Forget: Smarter Strategies for Summarizing Message History](https://blog.agentailor.com/posts/message-history-summarization-strategies.md): Explore effective strategies for summarizing message history in AI agents, ensuring they retain crucial context and information. - [Choosing Your Stack: LangChain & LangGraph in Python vs. TypeScript/JavaScript?](https://blog.agentailor.com/posts/langchain-vs-langchainjs.md): Side-by-side comparison of LangChain and LangGraph in Python vs TypeScript/JavaScript—APIs, agent patterns, ecosystem maturity, deployment, adoption stats, and practical guidance on when to choose each. - [How I Built and Deployed a Production-Ready AI SaaS in 14 Days Using Agent Initializr](https://blog.agentailor.com/posts/how-i-built-a-fullstack-ai-saas-product.md): A detailed case study on building a production-ready AI SaaS backend in under 14 days. I’ll show you how I used my own tool, Agent Initializr, to skip weeks of boilerplate and focus only on the features that mattered. - [GPT-5 Is Here — And It’s Built for Devs Who Build with Tools](https://blog.agentailor.com/posts/openai-gpt-gpt-5-release.md): OpenAI just released GPT-5, its most capable model yet. Here's what it means for developers building agents, tools, and AI-powered workflows — and how you can try it today using Agent Playground. - [OpenAI Releases GPT-OSS: What It Means for AI Developers and Agent Builders](https://blog.agentailor.com/posts/openai-gpt-oss-release.md): Explore OpenAI's new open-weight models GPT-OSS-120B and GPT-OSS-20B. Learn how they enable powerful agentic workflows, local inference, and safe open-source AI development. - [Prototype AI Agents with 1,000+ Tools in 2 Minutes — No Code, Full API Access](https://blog.agentailor.com/posts/prototype-ai-agents-with-agentailor.md): Launch fully functional AI agents with memory, tools, and APIs in minutes. Built for devs, indie hackers, and product builders. - [The Future of AI Building: Workflows, Agents, and Everything In Between](https://blog.agentailor.com/posts/agents-vs-workflows.md): Understanding the difference between AI workflows, agentic workflows, and true agents is crucial for building effective AI systems. This article breaks down these concepts and helps you choose the right approach for your AI projects. - [Run Any MCP Server Locally with Docker’s MCP Catalog and Toolkit](https://blog.agentailor.com/posts/docker-mcp-catalog-and-toolkit.md): Run MCP servers locally with Docker’s MCP Toolkit. Connect OpenAI agents to Notion, GitHub, Stripe and more — fast, secure, and containerized. - [Run Open-Source AI Models Locally with Docker Model Runner](https://blog.agentailor.com/posts/docker-model-runner-gemma.md): Learn how to run AI models using Docker Model Runner with Gemma 3. This guide covers installation, configuration, and running your first model. - [Getting Started with OpenAI’s Agents SDK for TypeScript](https://blog.agentailor.com/posts/openai-agent-typescript-sdk.md): Learn how to build AI agents in JavaScript with OpenAI’s new Agents SDK for TypeScript. This guide explores key concepts like agents, tools, guardrails, handoffs, and human-in-the-loop features—plus working code examples to get started fast. - [How to Build a Fullstack AI Agent with LangGraphJS and NestJS](https://blog.agentailor.com/posts/fullstack-ai-agent-app-with-langgraphjs-and-nestjs.md): Learn how to build a production-ready AI agent backend using NestJS and LangGraph.js. This guide walks you through project architecture, real-time messaging, LLM integration, and how to scaffold your own backend in minutes from a ready-made template. - [LangGraph vs LlamaIndex Showdown: Who Makes AI Agents Easier in JavaScript?](https://blog.agentailor.com/posts/langgraph-vs-llamaindex-javascript.md): Explore the key differences between LangGraph.js and LlamaIndex.TS—two powerful JavaScript frameworks for building AI agents. Learn about their architecture, features, and how to choose the right one for your next AI project. ## Key Pages - [All Posts](https://blog.agentailor.com/posts): Browse all articles on building AI agents. - [Reports](https://blog.agentailor.com/reports): Data-backed answers to one question each. - [Tags](https://blog.agentailor.com/tags): Browse posts by topic. - [About](https://blog.agentailor.com/about): About the author, Ali Ibrahim. - [Cameron](https://blog.agentailor.com/cameron): Building a local-first AI finance agent, in public — one release at a time.