# Article Architect MCP > Article Architect forces your AI agent to write technical blog posts that read like engineering arguments, not Wikipedia summaries. It demands a debatable thesis, exposes real-world tradeoffs, requires code blocks as evidence, grounds claims in production metrics, and defines a concrete action for the reader. This MCP guarantees structured reasoning, turning simple documentation into high-authority content. ## Overview - **Category:** productivity - **Price:** Free - **Tags:** technical-writing, blog, article, structured-reasoning, decision-pivots, content-architecture ## Description Stop getting articles that just list steps. Your AI agent usually writes tutorials: "Step 1: Install X. Step 2: Configure Y." It sounds clean, but it tells you nothing about why you should care or when the system breaks. Article Architect forces a change in approach. Instead of describing what something *is*, it makes your agent take a clear position—a thesis that someone smart might actually disagree with. You'll get articles that argue points, backed up by actual performance metrics and real-world failures. It demands proof: code comparisons and production data are mandatory parts of the output. The result is content built for engineers who need to know not just how something works, but what it costs to run and when they should avoid using it. Connecting Article Architect through Vinkius gives your agent a rigorous framework that elevates your technical writing from filler to genuine thought leadership. ## Tools ### architect_article Forces the AI agent to build a technical article around five pivots: stating a debatable thesis, exposing tradeoffs, planning evidence code, grounding it in production experience, and defining reader action. ## Prompt Examples **Prompt:** ``` Write a technical blog post about React Server Components. ``` **Response:** ``` Verdict: NO_THESIS. 'About React Server Components' is a topic, not an argument. What POSITION are you taking? 'RSC eliminated 4 round trips from our checkout flow and cut payload by 62% — but it moved complexity to the resolver layer.' That's a thesis. Take a side. What did RSC fix, what did it break, and when should you NOT use it? ``` **Prompt:** ``` Thesis: tRPC cut type errors 89% and API surface 60%, but onboarding slowed 3x. Tradeoffs: steep learning curve, runtime overhead. Code: type-error diff, API surface comparison, onboarding metrics. Evidence: 14-service migration over 3 sprints. ``` **Response:** ``` Verdict: PUBLISH_READY. Thesis is debatable (speed vs onboarding cost). Tradeoffs are painful (3x onboarding). Code is evidence (type-error diffs, not hello-world). Production is concrete (14 services, 3 sprints). Architecture validated — write it. ``` **Prompt:** ``` Thesis: Kubernetes simplifies container orchestration. Tradeoffs: minor complexity overhead. Code: kubectl apply example. Evidence: widely used in production. ``` **Response:** ``` Multiple rejections. (1) NO_THESIS — 'simplifies container orchestration' is a marketing tagline, not a thesis. Compared to what? At what cost? (2) ONE_SIDED — 'minor complexity overhead' dismisses real pain. (3) ABSTRACT — kubectl apply is documentation, not evidence. (4) THEORETICAL — 'widely used in production' is generic. Fix all four. ``` ## Capabilities ### Asserting a Debatable Thesis Forces the generated article to take an opinionated stand, stating what the technology achieved and at what cost. ### Exposing System Tradeoffs Ensures the write-up details failure modes and sacrifices inherent in a specific technical approach. ### Generating Evidence Code Blocks Requires code examples to serve as proof, like before/after comparisons or failing tests, instead of simple boilerplate. ### Grounding Claims in Production Data Mandates the inclusion of concrete operational details, such as specific metrics drops or failure timelines. ### Defining Reader Action Items Concludes the article with a clear next step—a benchmark to run or a migration pattern to apply. ## Use Cases ### Comparing two database technologies A platform engineer needs to compare Postgres and DynamoDB for a new microservice. Using Article Architect, the agent must argue which is better by citing specific failure modes (e.g., eventual consistency issues) and running code that proves one performs better under high write load than the other. ### Documenting a complex migration pattern A developer writes about moving from an old queue system to Kafka. The MCP forces them to establish a thesis (e.g., 'Kafka improved throughput but complicated debugging'), show code diffs for the transformation logic, and provide the exact migration checklist needed. ### Writing post-mortem analysis After an outage, the team needs to write a public technical report. Instead of vague descriptions, Article Architect ensures the article focuses on the root cause (the thesis), details why detection failed (the tradeoff), and provides concrete steps for prevention. ### Evaluating a new framework feature A team wants to publish about a new web component library. Using this MCP, they must write an article that doesn't just list features but argues its value by comparing it against older methods and showing performance metrics. ## Benefits - Stop publishing 'how-to' guides that lack punch. By mandating a debatable thesis, Article Architect forces the agent to argue a position—for example, stating how one service cut deploy time by 73%, even if it tripled complexity. - Your readers gain genuine value because the content exposes tradeoffs. The MCP ensures the article details when an approach fails or what you sacrifice, which builds immediate trust with skeptical engineers. - The code blocks aren't just syntax examples; they are evidence. Article Architect makes sure every code snippet serves a purpose, like showing before-and-after performance traces, not boilerplate hello-world code. - Claims get real weight by grounding them in production reality. Instead of 'works well,' the article must cite specific metrics, such as p95 dropping from 1.2s to 380ms after migration. - The final takeaway is actionable. The MCP ensures the piece doesn't just ask the reader to 'understand X.' It tells them exactly what to do next—like running a specific benchmark or using a new decision framework. ## How It Works The bottom line is you get content structured like an engineer's deep dive, not a marketing brochure. 1. You prompt your agent with a topic and tell it that the content must meet five strict architectural requirements: thesis, tradeoffs, code evidence, production data, and reader transformation. 2. The Article Architect MCP runs its validation checks. It will reject the structure if the argument is merely descriptive or if the code provided is generic documentation filler. 3. You receive a fully validated article blueprint that meets professional engineering standards for argumentation and proof. ## Frequently Asked Questions **How does Article Architect MCP guarantee the content isn't generic?** It forces five specific decision pivots: a debatable thesis, exposed tradeoffs, evidence code, production metrics, and reader transformation. If any one of these is missing or too vague, the tool rejects the output. **Can Article Architect MCP write about basic concepts?** No. The MCP is designed to fail if the article lacks an argument. It will reject anything that's merely a definition, forcing you to find an angle or a controversy to prove. **Do I need production data for architect_article?** Yes, it's mandatory. The tool requires at least one concrete operational detail—a failure metric or usage surprise—to ground the article in reality and give it authority. **How many tools are in Article Architect MCP?** It contains only one primary tool, architect_article. This single tool handles all five necessary structural validations for technical content creation. **What is the optimal input structure when using architect_article?** You must provide structured inputs covering all five decision pivots. Don't just submit a topic; supply preliminary drafts for the debatable thesis, potential tradeoffs, and concrete production metrics to start. **If architect_article rejects my output, what does that mean?** A rejection means your argument fails one or more structural requirements. The tool pinpoints the failure—whether it's a weak thesis or boilerplate code—so you know exactly where to strengthen your content. **Can architect_article handle non-software technical topics?** Yes, the pivots are universal and language-agnostic. While ideal for code, you can apply it to any complex claim by replacing 'code evidence' with a decision framework or compliance checklist. **Are there rate limits when running architect_article?** Vinkius manages the underlying resources for stability. We recommend pacing your requests, especially when processing multiple large articles, to ensure consistent quality across all runs. **Does Article Architect write the article?** No. Article Architect generates zero content. It forces the AI agent to architect the article's argumentative structure — thesis, tradeoffs, code strategy, production evidence, reader transformation — before writing a single paragraph. The architecture then guides the writing. The tool validates structural depth, not prose quality. **What is the difference between a thesis and a description?** A description tells you what something IS: 'Docker is a containerization platform.' A thesis takes a POSITION: 'Docker Compose replaced our Kubernetes cluster for 8 of 12 services — saving $4,200/month in infrastructure but requiring manual rollback procedures we hadn't planned for.' If a smart engineer could disagree with it, you have a thesis. **Why does it reject boilerplate code?** Code in a technical article is EVIDENCE, not illustration. If the reader can find the same code in the official documentation, the code adds nothing. Every code block must support the thesis: a before/after comparison showing the improvement, a benchmark proving the claim, a failing test that passes after the fix, a production trace revealing the bottleneck. Boilerplate teaches syntax. Evidence proves arguments.