Why I Do Not Use Chat Logs as My Knowledge Base
Explains why chat logs are not a durable knowledge base and how Obsidian, local Markdown, Codex, project directories, and publication nodes form a federated system.
codex
26 public items tagged codex.
What Obsidian Actually Does in an AI Knowledge Base
Explains Obsidian's real role in an AI knowledge base: local Markdown, links, backlinks, and human review, not an automatic truth machine.
How Codex Can Drive Verifiable SketchUp Modeling
Shows how an agent CLI, MCP server, SketchUp Ruby bridge, runtime skills, and a structured design model turn design intent into verifiable project state.
Why Codex Should Be a Worker, Not the Scheduler
Explains why Codex and similar AI coding tools should remain bounded workers, not long-running schedulers or release controllers.
Evidence Gate: Why AI Pull Requests Need Evidence Bundles
Explains why AI pull requests should include reviewable evidence bundles, not just code changes and a claim that tests pass.
An Issue Is Not a Todo: It Should Be an Executable Contract for AI
Argues that an issue should be an executable contract for AI work, with scope, context, gates, evidence, and release boundaries instead of a loose todo.
From Triage to Implementation: Why AI Fixes Need Gates
Explains why AI repair work should move through triage, analysis, implementation, evidence, and review gates instead of jumping straight to code.
Private drafts, public artifacts
A note on separating raw material from publishable work when Codex is part of the workflow.
Codex workbench
A local-first workflow for turning project folders, notes, scripts, and code into durable outcomes.
From Evidence to Wiki to Public Article
Shows how knowledge moves from evidence to wiki, then through publication source, prepare, review, and sync into a traceable public article.
kb lookup / promote / publish: Narrow Knowledge Interfaces
Shows how kb status, lookup, promote, and publish form narrow interfaces that let AI reuse, promote, and publish knowledge without crossing boundaries.
A Practical Domestic Alternative to Codex CLI and Claude Code
A practical comparison and installation guide for Codex CLI, OpenCode + MiniMax-M2.7, and Qwen Code + MiniMax-M2.7 as local AI coding CLI workflows.
Your Personal AI Workbench
You do not need to become an AI expert first, and you do not need to learn a pile of jargon. What ordinary people really need to practice is describing goals, boundaries, and acceptance criteria clearly. One command-line window where you type plain English and an AI does the work.
Step 1: Start Codex in an empty folder
The first step should not be folder design. Create one empty folder, start Codex there, and let Codex choose the first project structure.
Step 2: Open the first version Codex made
After Codex creates the first personal website in an empty folder, focus on the local preview, clear pages, and privacy risks.
Step 3: Publish a free public URL first
After the local site works, let Codex guide the GitHub and Cloudflare Pages authorization flow with copy-paste prompts instead of manual upload.
Step 4: Add a custom domain only when it helps
After the free URL works, ask Codex whether a custom domain is worth buying. You confirm accounts, domain choices, and page states; you do not need to learn DNS first.
Step 5: Keep editing the site after it is online
After launch, change one thing at a time: ask Codex for a plan, preview locally, check, and publish with the workflow this project actually uses.
Step 6: Check privacy and common mistakes before publishing
Once your personal site can be published and updated, copy one fixed pre-publish prompt so Codex checks privacy, images, links, and phone readability before every release.
Your first personal website with Codex, without fearing the tools
A plain-language entry point for people who use phones and computers but do not yet know hosting, GitHub, or DNS: start from one clean folder and let Codex guide the route.
My Minimum AI Workbench
Wrapping up the series with a copyable personal workflow configuration. One stable computer, one project directory, one command-line AI tool, one confirmation habit — this is the foundation for ordinary people using AI for real work.
Ten Things Ordinary People Can Do with AI Right Away
Ground abstract capabilities into real life and work. Ten categories of things ordinary people can complete with AI today — each with a concrete scenario and a copyable prompt.
Why I Think One Command-Line Window Is All You Need
CLI is not advanced — it is the cleanest entry point: no buttons to navigate, just one question — what do you want to do? Ordinary people do not need to learn a full software workflow first. They just need to state the goal and let AI guide them through the rest.
Why Ordinary People Can Use CLI Too
We start not from cd, npm, or git theory, but from one real scene: create a folder, open Codex, type one complete goal. Codex will guide you through the entire process. You only need to confirm at key moments.
Build your own independent site with Codex
A beginner-friendly guide for creators, families, engineers, and non-programmers who want a durable home on the web.
How much does a personal independent site really cost?
A practical cost breakdown for Codex, domains, static hosting, storage, email, and the upgrades that may come later.