Blog

Stateless MCP Architecture Part 03 | MCP 2026 Migration Guide: From Stateful to Stateless Without Just Updating the SDK

MCP 2026-07-28 is a protocol-era migration, not a package bump. This guide covers modern-first negotiation, dual-era serving, session and state migration, authorisation hardening, CIMD and DCR, and the correct deprecation paths for Roots, Sampling, Logging, and legacy HTTP+SSE.

Stateless MCP Architecture Part 02 | How Does MCP Stay Interactive Without Sessions? MRTR, Tasks, Headers, and Caching

MCP 2026-07-28 removes the long-lived protocol session, then separates multi-round input, long-running work, gateway routing, caching, and tracing into distinct mechanisms. This article explains MRTR, requestState, Tasks, HTTP metadata, ttlMs/cacheScope, and Trace Context by responsibility and failure boundary.

Moving to Codex and the Product Map Part 04 | ChatGPT Chat, Work, or Codex: Which Should You Open After July 2026?

Do not choose a ChatGPT surface from a feature list. Start with the evidence you would need before you could call the work done.

LLM Wiki: Theory part02 | How Does an LLM Wiki Work? From Markdown and Google OKF to Obsidian and Knowledge Graphs

How does an LLM Wiki actually run? Seven layers cracked open — raw sources, Markdown, OKF, links and the graph, search indexes, the Harness — plus how Obsidian and Sigma.js serve humans and AI from the same substrate.

Harness Theory Part 12 | Patterns, anti-patterns and evolution: how to grow a harness and know when to remove it

From behaviour-to-harness derivation, deterministic-first control, Build/Buy/Compose and adoption operating modes to entropy, knowledge governance, assumption ledgers, ablation and retirement, this article develops a method for harnesses that can grow, prove their value and become safely simpler.

Harness Theory Part 11 | Observability and operations: how do you know what the agent actually did?

From runtime and process observability, correlation hierarchies, telemetry schemas, traces, metrics and replay, to SLOs, incident learning, version correlation and human attention budgets, this article builds an operational evidence chain for agent systems.

Harness Theory Part 09 | Safety and Permissions: Letting Agents Act Without Letting Them Cross the Line

From agentic threat models, trust boundaries, permission precedence, human approval and sandboxing to credential binding, tenant isolation, data egress, extension supply chains and red teaming, this article explains how an agent harness confines autonomous action to authorised, isolated and auditable boundaries.

Harness Theory Part 08 | Reliability engineering: retries, fallbacks and side-effect control

From error classification, recovery state machines, retry budgets, idempotency, unknown outcomes and sagas to backpressure, asynchronous lifecycles, context degradation and scheduled delivery, this article explains how an agent harness recovers safely after failure.

Harness Theory Part 06 | Runtime and state: how does a harness sustain an execution world?

From initialisation, prompt assembly, model adapters and canonical run state to checkpoints, control planes, hosting, configuration versioning and protocol negotiation, this article examines who owns the truth after an agent starts work.

Codex Extensibility and Tooling Part 04 | Tool Calling vs Tool Search

Tool Search changes when a tool definition enters context; Tool Calling still handles invocation. The important threshold is schema cost, not a magic tool count.

Harness Theory Part 04 | Context Engineering: Where Do Context, Memory, and RAG Belong?

A practical guide to context builders, progressive disclosure, managed artefacts, compaction, memory, RAG, cache invalidation, and fresh-session reconstruction.

Harness Theory Part 03 | Agent Loops and Control Flow: How Does the Model Start Doing Work?

A practical account of tool calling, agent-loop invariants, workflows and orchestration, item–turn–thread protocols, bidirectional control, stopping policy, and goal drift.

Harness Theory Part 01 | What Is a Harness, and Why Do Agents Need One?

A practical definition of an agent harness, its boundary with the model and product, harnessability, and the human steering control plane.

Codex Daily Workflow and Controls Part 06 | Record & Replay: Turn a Demonstrated Workflow into a Reusable Skill

Some SOPs are easy to perform and tedious to specify. Record & Replay lets Codex observe one run, draft an editable Skill, and reuse the procedure with different inputs.

Codex Agentic Engineering Part 02 | Context Canvas for Long-Running Tasks

Long Codex tasks often waste time after compaction by reopening files and rebuilding decisions that were already settled. Context Canvas keeps a small task map outside the conversation, points back to durable evidence, and gives the agent a route for resuming without rediscovering the project.

Codex Daily Workflow and Controls Part 05 | Keeping Codex Working When You're Away from Your Desk

Codex does not need a second development environment on your phone. This guide shows when to use Appshots, Remote, SSH, Handoff and locked Computer Use, and when the local path is already enough.

The Atlas of Agent Design Patterns Part 9 | How to Choose an Agent Architecture

A rigorous decision process for choosing the smallest agent architecture that satisfies task dynamics, evidence, risk, authority, recovery, memory, cost, and operational requirements.

Codex Multi-Agent and Model Orchestration Part 02 | How Should the Lead Agent, Workers, and Checkers Split the Work?

A permissions refactor shows where a Lead Agent, Workers and Checkers need clear boundaries around decisions, writable state, evidence and final acceptance.

The Atlas of Agent Design Patterns Part 7 | Context, State, Memory, and RAG

A production-focused guide to working, episodic, semantic, procedural, user, and shared memory; the boundary between memory and RAG; memory write and retrieval policies; conflict resolution; forgetting; and production memory architecture.

Codex Daily Workflow and Controls Part 04 | Goal Mode: Letting Codex Work for Hours Without Losing Control

Goal Mode suits long jobs with a stable destination. This article uses an auth migration to show how to define done, leave evidence as the work progresses, and decide when to steer or replace the Goal.

The Atlas of Agent Design Patterns Part 5 | Verification, Recovery, and Self-Correction

A production-focused guide to retry, parameter repair, fallback, Self-Refine, critics, verifiers, Generate-and-Test, replanning, Reflexion, human approval, idempotency, and bounded recovery.

Codex Git Workflows & Collaboration Part 02 | What to Check After Codex Says 'Done': Diffs, Reviews, Pull Requests and Evidence

A practical review workflow for Codex changes: identify the exact change set, inspect what actually ran against it, then apply review and pull-request policy before merging.

The Atlas of Agent Design Patterns Part 3 | ReAct, Plan-and-Execute, Adaptive Planning, and HTN

A production-focused guide to fixed decision logic, bounded ReAct, explicit plans, adaptive replanning, hierarchical decomposition, HTN, plan verification, and the architecture that combines Planner, Executor, Verifier, State Machine, and policy controls.

The Atlas of Agent Design Patterns Part 2 | Agent Execution Paths: Direct Calls, Pipelines, Routers, State Machines, and DAGs

A practical guide to the execution structures behind production agent systems, including direct calls, fixed pipelines, routers, state machines, DAGs, event-driven triggers, human approval, and behavior trees.

The Atlas of Agent Design Patterns Part 1 | Beyond ReAct: A Six-Dimensional Map of LLM Agent Architectures

A practical engineering map for analysing LLM agent architectures across execution path, decision and planning, reasoning and search, verification and recovery, agent organisation, and state and memory.

Codex Daily Workflow and Controls Part 02 | Scheduled Tasks, Cron, and What the Agent Still Has to Decide

A practical way to separate scheduling from agent execution, choose Local or Worktree, avoid unnecessary model pinning, and verify that a recurring task actually completed.

Codex Git Workflows & Collaboration Part 01 | Git, Branches, Worktrees and Pull Requests: A Mental Model for Codex Multi-Agent Work

Branches carry lines of development, worktrees give concurrent tasks separate working directories, and pull requests organise review and integration. This article builds a practical Git model for multi-agent Codex work.

Codex Daily Workflow and Controls Part 01 | Plan, Steering, and Choosing Where to Control and Run Codex

When Codex drifts halfway through a task, should you steer it, queue the next request, fork the work, or stop the turn? This guide starts from those real supervision decisions and separates the interface you use from the environment doing the work.

Codex Extensibility and Tooling Part 03 | Hooks: Run Logic on Lifecycle Events

Putting a validator on Stop is different from telling a Skill to run tests before finishing. Event delivery and trust decide whether the Hook actually runs; Skills, MCP and Plugins handle different parts of the setup.

Moving to Codex and the Product Map Part 03 | Codex Is More Than a Model: Understanding the Agent Harness and Command Center

A practical explanation of the system around the model in Codex: the harness, worktrees, permissions, Hooks, feedback and human supervision.

From Pain to Venture Part 11 | From product to venture system: operations, resources, team, finance, and risk

The product is the part customers see.

From Pain to Venture Part 10 | Demand creation: having a product does not mean people will come.

Many products do not die because they have no value.

From Pain to Venture Part 09 | Market, competition, and positioning: you are not building in a vacuum

The earlier parts looked at problems, customers, MVPs, and business models.

From Pain to Venture Part 07 | An MVP is not a smaller product. It is a learning machine.

When people say MVP, they often mean “a cheaper version of the product”.

From Pain to Venture Part 06 | From problem to solution: do not invent the product yet, invent the hypotheses

Once a problem has been found, it is tempting to relax.

Codex Extensibility and Tooling Part 02 | Auto-review: Who Decides When Codex Can Cross the Sandbox Boundary?

Codex can send eligible sandbox-boundary approvals to a separate reviewer agent without removing the sandbox. OpenAI's 99.1% figure is an approval rate, not a security score.

From Pain to Venture Part 04 | A pain point is not a complaint. The useful ones are important and still unmet.

Founders are easily drawn towards complaints.

From Pain to Venture Part 02 | Tools are not for memorising. They are for forcing the problem into focus.

The most dangerous thing about tools is not that they are useless.

From Pain to Venture Part 01 | Do not start with the idea. Start by admitting where things are stuck.

Some venture ideas are not born in front of a whiteboard.

x402: Invisible Web3 02 | x402 Is Not Stripe for AI Agents. It Is a Paid Request Grammar

x402 is often described as Stripe for AI agents, but a more precise framing is paid request grammar: a way for HTTP requests to quote, authorise, pay, verify and deliver.

x402: Invisible Web3 01 | The Most Successful Web3 May Be the One Users Never Notice

An opinion piece on invisible Web3, x402, AI agent payments, and why mature infrastructure should help users complete tasks instead of forcing them to learn the rails.

Product Rhythms in the Age of AI Part04 | If I Were Running a Product Team Today, I Wouldn’t Run Scrum the Same Way

Product Rhythms in the Age of AI Part03 | I’m Less Interested in Whether We “Use Scrum” and More in Which Rituals Still Earn Their Keep

Product Rhythms in the Age of AI Part02 | AI Won’t Replace PMs First. It Makes Some Work Cheaper Before It Does Anything Else

Product Rhythms in the Age of AI Part01 | It’s Not Agile That’s Aging. It’s Some of the Ways We Deliver

Business Models, Revenue Engines & Unit Economics Part 03 | How Can the Same Product Become a Different Business With a Different Pricing Model? Pricing, Packaging, and Monetization Explained

The same product can behave like a different business under subscription, per-seat, usage, committed-use or outcome pricing. A framework for pricing metrics, packaging, risk allocation and unit economics.

What Reality Corrected 08 | Culture is not what you write on the wall. It is what your decisions repeatedly train people to do

Before Jevan Soo Lenox’s fireside at Draper U started, I had a fairly defensive posture towards the topic. Anything l…

What Reality Corrected 07 | Early startup value is not your title. It is how quickly you turn ambiguity into useful work

At Draper U, when Cristina Cordova started talking about her path through Stripe, Notion, First Round and now Linear…

What Reality Corrected 06 | A better product is not enough if switching still hurts

In another Draper U session, the speaker asked a nastier question than “why would a customer buy?” **If the customer …

Codex Extensibility and Tooling Part 01 | Skills, Plugins, and MCP: What Does Each One Actually Do?

A practical guide to Skills, MCP, Plugins and Tool Search: where to keep repeatable working methods, how external capabilities enter a workflow, how large tool catalogues stay searchable, and when packaging becomes worth the effort.

Business Models, Revenue Engines & Unit Economics Part 01 | Where Does Revenue Come From, and Why Does It Change? A Practical Revenue Architecture Framework

Do not stop at the revenue growth rate. Build a Revenue Architecture that connects earning mechanisms, activity denominators, commercial metrics, recognition boundaries and a reconciled revenue bridge.

What Reality Corrected 03 | A pitch deck is not your company story. It is the thing that decides whether an investor keeps reading

At Draper U, Melanie opened her pitch workshop with a line that was more useful than any deck template. Imagine your …

What Reality Corrected 02 | It wasn't that you were bad at networking. Most networking events are just badly designed

The first time this really clicked for me was not in class. It was after I hosted a founder × investor mixer in the B…

Local LLM Fine-Tuning, Explained Appendix B | Commands, warnings and troubleshooting quick reference

Financial Fundamentals & Financial Statement Interpretation Part 03 | Why Can Revenue Grow Without Making the Business Better? From Revenue Quality to Operating Leverage

Revenue growth is only the starting point. This article builds a four-layer framework for judging whether business economics are genuinely improving, using revenue quality, margin boundaries, cost behaviour, operating leverage and revenue bridges.

Local LLM Fine-Tuning, Explained part 09 | How to choose a mainline version: Lexi, adapter variants and the judgement that remains

what should I actually keep?

Local LLM Fine-Tuning, Explained part 08 | From adapters back to Ollama: merge, quantisation and deployment

a finished training run is not the same thing as a finished model

Financial Fundamentals & Financial Statement Interpretation Part 02 | What Happens Between Profit and Cash? Understanding Cash Flow Through Working Capital

Why can a profitable company still lose cash? Trace the gap from accrual profit to operating cash flow, working capital and liquidity without mistaking reconciliation for diagnosis.

Local LLM Fine-Tuning, Explained part 06 | How to read the training scripts: from `train_lora.py` to `train_partial_ft.py`

an experiment design document that binds together the model, the data, the tokenizer, the training method, the trainable scope and the training rhythm

Local LLM Fine-Tuning, Explained part 04 | What SFT, LoRA and full fine-tuning each actually change

SFT answers how you teach. LoRA answers how you change parameters. Full fine-tuning answers how deep you are willing to cut.

Local LLM Fine-Tuning, from Modelfile and LoRA to DPO part 03 | How the toolchain splits: Hugging Face, Transformers, PEFT, TRL, and Ollama

Financial Fundamentals & Financial Statement Interpretation Part 01 | Why Can Revenue Rise Before Cash Arrives? Understanding the Three Financial Statements Through One Credit Sale

Follow one transaction from recognition to the balance sheet and cash settlement, then use the remaining gap to ask a better business question.

Local LLM Fine-Tuning, from Modelfile and LoRA to DPO part 01 | The layer map of a local LLM: from prompts and Modelfiles to model weights

Business & Financial Analysis | A Complete Map from Company Numbers to Management Decisions

How financial statements, business models, plans, forecasts, liquidity, KPIs and variance diagnosis form one repeatable decision chain.

LLM Application Engineering and Governance for PMs Part 3 – LLMs Do Not Do the Work Alone: How Tool Calling and RAG Connect Models Back to the Real World

The first instinct many teams have when they start building with LLMs is fairly predictable: write a better prompt, add a few examples, and see whether the model can somehow hold the whole thing toget

LLM Application Engineering and Governance for PMs Part 1 － Beyond Prompt Engineering: What PMs Are Actually Doing When They Put LLMs into Products and Workflows

- You already use ChatGPT, but you’re now trying to wire LLMs into a product, an internal workflow, or some form of automation - You keep hearing prompt engineering, RAG, tool calling, agents, and it

Category Design for Founders Part 2 － How Founders Actually Do Category Design: Problem Framing, POV, Languaging, Fundraising, and Go-to-Market

Part 1 dealt with one thing first: category design is not a naming game, and it is not a grand branding wager every startup should force. At its core, it is about correcting a market that is using the

After the Pause | On Starting, Pivoting, and Pausing a Venture Over Two Years — Part 07: Going from founder back to candidate was harder than I expected

I assumed one thing would at least become simpler once I came back to the job market after a stretch of building a company.

After the Pause | On Starting, Pivoting, and Pausing a Venture Over Two Years — Part 06: I Did Not Give Up. I Just Needed to Stay Afloat

In the end, I did not pause the venture because I stopped believing in the problem.

After the Pause | On Starting, Pivoting, and Pausing a Venture Over Two Years — Part 04: I Thought Product Would Be the Hard Part

At the beginning of my second venture, I placed most of my attention where many founders do.

After the Pause | On Starting, Pivoting, and Pausing a Venture Over Two Years — Part 03: It Was Not a Lack of Direction. Reality Kept Changing the Question

There was a stretch when my pitch deck changed every few weeks.

After the Pause | On Starting, Pivoting, and Pausing a Venture Over Two Years — Part 01: I Did Not Suddenly Decide to Start a Company

One of the most common reactions I get, once people learn I went off to build a start-up, goes something like this.

PM User Research and Fieldwork 09 － A Complaint Is Not Yet a Job: How PMs Confirm JTBD, Switching Moments, and Pain Intensity Without Jumping to Solutions

A complaint is raw input, not yet a problem definition. This article shows how to distinguish symptoms, needs, jobs, constraints, and actionable opportunities.

PM User Research and Fieldwork 07 － Interviews and Fieldwork Are Not Casual Conversations: How PMs Facilitate, Observe, Take Notes, and Avoid Distorting the Session

Good interviews and fieldwork require disciplined facilitation. This article covers rapport, neutrality, probing, observation, and session control.

PM User Research and Fieldwork 06 － An Interview Guide Is Not a List of Questions: How PMs Design Non-Leading Prompts That Surface Real Evidence

An interview guide is a decision tool, not a list of questions. This article shows how to structure prompts, probes, and follow-ups without leading participants.

PM User Research and Fieldwork 04 － Five Participants Is Not a Recruitment Strategy: How PMs Find the Right Users, Screen Them Properly, and Avoid the Wrong Ones

Recruiting five participants is not a strategy. This article explains how research goals, segment variation, risk, and saturation should shape participant selection.

PM User Research and Fieldwork 03 － Not Every Research Session Is an Interview: User Interviews, Usability Tests, Field Studies, and Diary Studies Serve Different Jobs

Interviews are only one research method. This article compares interviews, observation, diary studies, usability tests, and other approaches by the evidence they produce.

PM User Research and Fieldwork 01 － When the Dashboard Speaks but the User Does Not: Why PMs Need User Research to Close Product Blind Spots

Dashboards show what happened, but not always why. This article explains when product managers need user research to recover the missing context behind the metrics.

PM Growth Levers and Monetisation 06 － Growth OS: Mature PMs Do Not Just Generate Ideas—They Turn Metrics, Experiments, Decisions, and Scale into a Repeatable Operating System

A surprising number of teams are not short of ideas. What they are short of is rhythm.

PM Growth Levers and Monetisation 04 － Monetisation Is More Than Slapping on a Paywall: How to Choose Between Fake Doors, Trials, Bundling, and Usage Limits

When teams start talking about monetisation, the conversation often rushes straight to the paywall.

PM Growth Levers and Monetisation 03 － Lifecycle Is Not About Sending More Nudges: Retention Is About Designing a Reason to Return

When teams say they want to improve lifecycle, what they often mean is “we should probably send more messages”.

PM Growth Levers and Monetisation 01 － Activation Is Not Onboarding: PMs Should Start with the Aha Moment, TTFV, and the First Verifiable Value Moment

A lot of teams say they want to improve activation, but what they actually end up working on is onboarding.

PM Product Data and Experimentation 08 － Data Rarely Breaks at the SQL Layer: PMs More Often Fight Identity Issues, Late Events, Bot Traffic, and Failed Rollouts

When PMs see strange numbers, the first instinct is often something like this:

Codex Multi-Agent and Model Orchestration Part 01 | Spark and Mini Subagents: Speed, Cost, and First-Generation Model Specialisation

A coding task can spend time searching, making design judgements, or simply waiting for the next visible edit. GPT-5.4, GPT-5.4 mini and Codex-Spark make this early Codex model split unusually easy to see.

PM Product Data and Experimentation 06 － Retention, Cohorts, and Segmentation: Knowing Who Stays Matters More Than Watching the Average

A product can look healthier than it really is simply because averages are excellent at cosmetics.

PM Product Data and Experimentation 04 － SQL for PMs: Learn to Pull Evidence Before You Claim to Be Data-Driven

When PMs say they want to become more data-driven, the first instinct is often to ask for more dashboards.

PM Product Data and Experimentation 03 － A Tracking Plan Is Not an Event Checklist: It Is a Data Contract Between PM, Engineering, and Analytics

When teams talk about a tracking plan, they often sound as if they are managing a to-do list.

PM Product Data and Experimentation 01 － North Star Is Not Enough: How PMs Use HEART, Guardrails, and Counter-Metrics to Define Success

When PMs first get introduced to product metrics, they usually hear a powerful idea very early on: find your North Star Metric, and the whole team will start pulling in the same direction.

RAG Engineering in Practice 07 – How My Job Agent Turns JDs, CVs, and Rubrics into a Working Evidence Pipeline

If the first six articles in this series were mostly about sharpening concepts, setting boundaries, and clearing technical debris off the road, this one is about putting those ideas back into a real workflow and seeing how they actually hold together.

RAG Engineering in Practice 05 – Qdrant’s JSON 400 Hell: The Problem Is Usually Not Qdrant but the Body You Actually Sent

If you are calling Qdrant from Make or another low-code tool, the thing most likely to drive you up the wall is usually not vector retrieval itself. It is the class of JSON 400s that look utterly unreasonable.

RAG Engineering in Practice 04 – Why Qdrant Throws `Index required`: Payload Indexes, Schema, and Filter Design Explained

The first time you add a filter in Qdrant, there is a decent chance you will run into an error that feels oddly rude:

RAG Engineering in Practice 02 – From Relational to Vector: A Practical Database Selection Map

A lot of database discussions still begin with “SQL versus NoSQL”. That is not wrong, exactly. It is just usually too coarse to be very useful for real engineering decisions in 2026.

RAG Engineering in Practice 01 – RAG Is More Than a Vector Database: Start with the Full System Map

When people first get into RAG, their attention often goes straight to the vector database. That is understandable. It is usually the part that sounds the most new, the most technical, and the most likely to impress someone in a meeting.

ComfyUI Series 06 | How do you install models? Checkpoints, clips, loras, vae, T5XXL, GGUF, and FP8 explained

This is the point where things stop being “I vaguely understand the theory” and become “why is my workflow full of angry red boxes”.

ComfyUI Series 05 | How do you actually choose a model? SD 1.5, SDXL, LCM, Turbo, Pony, Flux, and HiDream explained

Once ComfyUI is up and running, the next thing that tends to melt people's brains is the model list.

ComfyUI Series 03 | Don’t just click about blindly: `127.0.0.1:8188`, your first image, tmux, background runs, and what happens after a reboot

Who is this for? Anyone who has managed to install ComfyUI, but is now staring at the interface wondering what on earth to do next. If you haven’t installed it yet, read the previous piece first. This one assumes ComfyUI already launches on your Mac.

ComfyUI Series 02 | Installing ComfyUI on a Mac mini M4: get Python, PyTorch and MPS sorted first, and life gets easier later

Who is this for? Anyone who has decided to use ComfyUI and now wants a clean, workable install on a Mac mini M4 with 16GB. We’re not doing the full model menagerie yet, and we’re definitely not diving into Flux on day one. This piece is about laying a sane foundation.

Building AI Skills Series Part 3: Designing a skill layer from existing tools, using a Make-based job workflow as the example

A practical method for designing a skill layer above existing APIs, automations, and Make workflows without rebuilding the execution stack.

Building AI Skills Series Part 2: Why ChatGPT should not connect directly to low-level tools

Why exposing low-level tools directly to ChatGPT becomes difficult to govern, and how a thin skill gateway creates safer task boundaries.

MCP Engineering Deep Dive 04: Skills Are Not the Same Thing as Runtime Tool Selection

Having skills in your repo does not mean the runtime will actually route by them. In practice, the model is far more influenced by the tool surface you expose, the live inventory a session can see, and whether your descriptions, schemas, and instructions line up.

Moving to Codex and the Product Map Part 01 | Codex in 2026: From Coding Agent to an Agent Work OS

A dated map of how Codex expanded through parallel agents, computer use, goals, remote work, Record & Replay, knowledge work, Multi-Agent V2 and Agent Plugins during 2026.

MCP Engineering Deep Dive 02: Security, Auth, and Public Server Hardening Are Product Boundaries, Not Patches

If your MCP server is public, can write data, or touches real users, security cannot be the last middleware you add. It has to shape your tool surface, auth flow, authorization scope, and response discipline from the beginning.

MCP Engineering Deep Dive 01: Transport and Remote Deployment Are Part of the Design

Part 01 of the MCP Engineering Deep Dive series.

Build Your Own MCP Server — Part 4: What Skills Are, and How to Design a Skill Layer

Part 4 of the MCP Server build series.

Build Your Own MCP Server — Part 3: How to Compare MCP Frameworks and Choose One

FastMCP, MCP Framework, xmcp, and Spring AI MCP Server are not four skins over the same idea. The real question is which one fits your language habitat, deployment model, auth pressure, and long-term operating style.

AI Agentic Workflow Series 7 — How AI connects to the world: MCP will not replace workflows, it is the middle layer that lets agents use tools safely

MCP is not a workflow killer, and it is not merely a trendier wrapper around webhooks or function calling. Its real role is to expose tools, data, and prompts through a standardised interface so agents can connect to external systems more safely and consistently.

Build Your Own MCP Server — Part 1: What MCP Actually Changes

This is not just about making tools callable. It is about redrawing the line between reasoning, capability exposure, and execution.

AI Agentic Workflow Series 5 — Why I Eventually Replaced the Make MCP Server with My Own Gateway Server

v2 solved the “Make-first workflow as execution engine” problem. v3 takes the next step and moves the capability surface the model actually sees from the Make MCP server into a self-hosted FastMCP skill gateway.

AI Agentic Workflow Series 4 — Refactoring a Make-first workflow into an MCP execution engine: a practical migration from toolisation to contracts

How I migrated a Make-first job agent into a clean MCP execution engine, moving from flow-driven design to contract-driven architecture.

AI Agentic Workflow Series 2 — How to Build a Context-Aware Job Agent in Make — Part 1

Part 1 of a practical build log on creating a context-aware job agent in Make, covering LINE intake, continuation handling, routing, and recent-job scraping.

OpenClaw Deployment and Configuration Part4

When OpenClaw breaks, how to sort the failure into the correct fault line before you waste hours guessing.

OpenClaw Deployment and Configuration Part3

How to handle permissions, updates, backups, and migration without gradually wrecking a working OpenClaw host.

OpenClaw Deployment and Configuration Part2

After installation, which settings are worth touching first, and which ones should wait.

OpenClaw Getting Started 07 | Can a Mac mini M4 with 16GB RAM Run a Local LLM? Why I Started with Ollama and Llama 3.1 8B

If you have a **Mac mini M4 with 16GB of unified memory and a 256GB SSD**, the first problem with local models is rarely installation. It is choosing a sensible starting point.

OpenClaw Getting Started 06 | How to Choose Models: OpenAI, Claude, Gemini, Grok, MiniMax, z.ai, and local models

A practical model-selection guide covering cost, latency, reasoning, tool use, and my real OpenClaw setup.

OpenClaw Getting Started 04 | How to Use OpenClaw in Your First Week: The Right Order for CLI, Dashboard, Memory, Browser, and Discord

A practical first-week operating guide based on my macOS and Codex OAuth setup.

OpenClaw Getting Started 03 | Discord, Skills, Memory, Remote Access, and the Safety Baseline

For readers who already have the base install working and now want a usable but controlled OpenClaw setup with Discord, skills, memory, remote access, and safer execution boundaries.

OpenClaw Getting Started 01 | What OpenClaw Actually Is: A Self-Hosted Agent Gateway, Not a Stronger Chat UI

A technical introduction for readers who are comfortable with the terminal and want a correct mental model of OpenClaw before installing it.