LEARNING PATH 02 / INEMA AREAS

Tools and environment

Claude → Codex, Codex + Claude, and OSWork: where agents work and how to avoid getting locked into just one.

0% 0 of 0
01 / LEARNING PATH 2Claude → Codex02 / LEARNING PATH 2Codex + Claude03 / LEARNING PATH 2OSWork04 / LEARNING PATH 2AREA WORKSHEET05 / LEARNING PATH 2FIRST STEP
Three areas, with the same deliverables for each: the worksheet and the first step.
3 areas
18 topics
60 min with practice
1 worksheet per area

Learning path map

2.1 ~20 min

🧭 Claude → Codex: Separate the brain from the model

migrate or stay tool-agnostic

2.2 ~20 min

🧭 Codex + Claude: one does the work, the other reviews it

use both together

2.3 ~20 min

🧭 OSWork: Which Edition Is for You?

from chat to your agent environment

Detailed content

MODULE 2.1

Claude → Codex: Separate the brain from the model

Explain why lock-in lives in the runtime and how the Audit → Adapt → Prove → Handoff method moves project knowledge into a portable core.

0% 0 of 0

What it is

The Claude → Codex area starts with one statement: “Don’t migrate your brain. Separate the brain from the model.” The “brain” is everything you’ve accumulated in a project: context, rules, decisions, tasks, skills, handoffs, and memory. According to the page, only CLAUDE.md and AGENTS.md are tied to a provider; everything else is portable Markdown, plain text that any agent can read. When that material lives in a portable layer inside the project, Claude Code, Codex CLI, Gemini, or a local model in a container become just executors: programs that do the work based on those files. Switching models becomes a change at the boundary, not at the center. The area brings together the thesis, the method, a course with 3 tracks, a kit of scripts, and mega-prompts that put it into practice.

Why learn it

Models change often because of price, access, or quality. If your project knowledge is in the right place, each switch costs little instead of requiring you to rebuild it.

Key concepts

brain = context, rules, decisions, tasks, skills, handoffs, and memory; portable Markdown; executor; change the boundary, not the center

What it is

Lock-in means being tied to a provider because leaving would be costly. The page states: “Lock-in isn’t in the model. It’s in what you’ve left inside the runtime.” The runtime is the program that runs the agent, such as Claude Code or Codex CLI. Someone who has used a coding assistant for a few months may have accumulated instructions, memory, skills, hooks, and thousands of sessions in a format only that runtime can read. The page describes three changes: lock-in is invisible until the day you switch; the same knowledge can work with any model; and you have less dependence and more control. The content lasts; the format is disposable. Mixing them is what makes migration harder.

Why learn it

This area is for people who use Claude Code or Codex on real projects and worry about switching models. Understanding where the dependence lives shows what needs to move out of the runtime before a switch becomes necessary.

Key concepts

lock-in; runtime; CLAUDE.md, native memory, and JSONL sessions store the work; they aren’t the work; durable content, disposable format

What it is

The method has four steps, in this order: Audit, Adapt, Prove, and Handoff, “never implementing before auditing.” Auditing means making a read-only inventory of both runtimes: skills, commands, subagents, hooks, and MCP. Each skill is classified as reusable, an adapter, native, or unresolved. Adapting means splitting the CLAUDE.md: portable rules go into AGENTS.md, while anything that depends on a Claude plugin, hook, or menu stays in the specific file, which imports only the portable rules. Proving means opening a new session and asking five continuity questions: the goal and success criteria, a rule and its source file, the last decision, the next action, and any conflicts. Handoff means recording decisions, open items, next steps, and paths in a Markdown file that the next session reads before acting.

Why learn it

This order avoids the costliest mistake: changing files before you know what’s there. The proof rule also keeps you from confusing a created file with knowledge the agent actually used.

Key concepts

Audit → Adapt → Prove → Handoff; MODE: audit; portable × residue; five continuity questions; file existence is not proof

What it is

The portable core has seven places, each with an owner and an update rule. AGENTS.md holds stable rules and the reading order; context/overview.md, verified facts with their sources and dates; context/current-state.md, what works and what’s pending; context/sources.md, where each piece of information comes from. context/decisions/ holds one accepted decision per file; tasks/current.md states the goal, owner, success criteria, and next action; handoffs/latest.md prepares the next session to continue. The page explains that these names are conventions: no runtime loads these folders on its own. The reading order at the top of AGENTS.md tells it to read them. Each piece of information has a type: fact, preference, hypothesis, or decision, and “provenance wins over timestamp,” meaning the source matters more than the date and nothing is deleted; use superseded_by instead.

Why learn it

Mixing facts with hypotheses, or decisions with preferences, can make the agent repeat old mistakes and contradict what’s already been resolved. Owners and dates let you promote information to approved context, always with your approval.

Key concepts

AGENTS.md; context/; decisions/; tasks/current.md; handoffs/latest.md; fact, preference, hypothesis, decision; provenance wins over timestamp; keep secrets out of the repository

What it is

The page breaks the decision into three effort levels, not opinions. Level 1 is to import in the app with one click: it’s quick, but limited to what the other side accepts. The Codex CLI, for example, has no native import. Level 2 is to migrate with the kit using one command: run the agente-claude-codex scripts to audit, adapt, install the core, port skills, and prove. Level 3 is a durable, portable personal layer: knowledge lives in the project, and the runtime is interchangeable. Staying model-agnostic, meaning you don’t depend on any specific model, is Level 3 and, according to the page, the only option that survives the next switch. Being model-agnostic doesn’t mean leaving Claude: you can still choose the executor for each task.

Why learn it

Not everyone needs Level 3 right now. The page lists when to migrate now, due to cost, access, or policy, and when it makes sense to stay model-agnostic, such as when you use two executors or work with client projects.

Key concepts

Level 1: import; Level 2: migrate with the kit; Level 3: portable personal layer; migrate now × stay model-agnostic; cost of doing nothing

What it is

The course Claude → Codex: migrate or stay model-agnostic is free and available in Portuguese, with English and Spanish versions. According to the curriculum, it has 3 tracks, 18 modules, and 108 topics. Track 1 covers fundamentals and vocabulary, Track 2 walks through the kit command by command, and Track 3 presents six projects based on a real system. The agente-claude-codex kit uses bash and Markdown, with three ways to use it: scripts, mega-prompts, and the portable core template. doctor.sh answers “is my environment ready?” and audit.sh saves a read-only report. The kit also includes adapt-instructions.sh, init-core.sh, sync-skills.sh with polyskill, readback-test.sh, drift-report.sh, and promover-memoria.sh. The page sums it up: “The best first step is one of your own projects, in audit mode.”

Why learn it

The course explains why, and the kit does the work without deleting anything: nothing in ~/.claude or ~/.codex is copied in bulk or deleted, and installing a skill creates a backup alongside it. Starting with the audit shows you in one session what’s portable and what depends on a hook.

Key concepts

3-track course; agente-claude-codex kit; doctor.sh; audit.sh; mega-prompts A and B; Topic Map; sister area Codex + Claude

View full version →

MODULE 2.2

Codex + Claude: one does the work, the other reviews it

Explain how to use Claude and Codex on the same project, covering the Use Both kit’s six levels, the routing card, and the difference between claudex and Use Both.

0% 0 of 0

What it is

The Codex + Claude area’s central idea is: “One plans, the other critiques. Use both together.” Claude and Codex complement each other when each handles one part and the other checks the real artifact: the actual file, diff, or image. The rule that runs through it all: the second assistant gets the same brief and the real artifact, never a vague summary of what the first assistant did. The page is INEMA’s meeting place for the two tools. It brings together the Use Both kit’s six levels, a table showing who does what, and cards for all courses and projects. To move from one tool to the other, use the sister area Claude → Codex.

Why learn it

This is for people who already have Claude Code and Codex and don’t want to choose just one. A second opinion from another model catches issues the author may miss.

Key concepts

one does the work, the other reviews it; real artifact; same brief; it’s not Claude or Codex; sister area Claude → Codex

What it is

The Use Both kit organizes working with both tools into six workflow levels, with ready-to-use prompts and two small skills. Level 1 is having them discuss: one writes the plan, the other critiques the same file in read-only mode, for no more than two rounds. Level 2, optional, is generating images, only if an image tool is available in the session and after checking access and billing. Level 3 is splitting building and review: one builds in a branch, the other reviews the diff against the brief and test results. Level 4 is supervised computer use, with passwords, payments, and important submissions kept under your control. Level 5 is a goal with a stop condition, observable success, and limits on files, rounds, and spending. Level 6 is handoff and prime: the handoff records the state in a portable file, and prime makes the next assistant read and verify that state before continuing.

Why learn it

The levels give you a path to follow: you can start by discussing a plan and move up only when you need to. Each level includes limits to prevent uncontrolled spending or actions.

Key concepts

discuss; generate images; build and review; supervised computer use; goal with a stop condition; handoff and prime

What it is

The route card is a table in the kit that says, for each task, who does the first pass, who does the second, and what counts as done. Plan the work: Claude / Opus plans, Codex critiques the assumptions, and the result is a plan with scope and acceptance tests. Build a feature: Claude / Opus builds, Codex reviews the diff, and done means tests pass and findings are resolved. Tackle a difficult bug: Codex works toward a bounded goal, with tests and a human checkpoint. Switch sessions or models: create a handoff snapshot, and prime checks the current state. The page notes that these are "editorial choices, not a measured model ranking": try them and keep the route that produces the best evidence.

Why learn it

Having a starting route saves you from debating from scratch for every task. Knowing it is not a benchmark keeps you from treating it as a fixed rule.

Key concepts

first pass; second pass; finish line; editorial choices, not a ranking; models and billing vary by account

What it is

The page organizes courses and projects into one path: understand, migrate or stay tool-agnostic, use both together, and automate the discussion. The common link is the project’s Markdown files, such as AGENTS.md, tasks, and handoffs, which any runtime can read. Two different tools appear along this path. claudex is a Claude Code plugin, or software: Claude writes PLAN.md and Codex critiques it from 3 angles, in a loop until LGTM or N rounds. Use Both is a method kit, with a guide, prompts, templates, and 2 skills. It covers all six workflows, manually or semiautomatically. The page sums it up: they are complementary; for a large plan, use claudex; for everything else, such as reviewing a diff, setting a goal with a stop condition, or switching sessions, use Use Both.

Why learn it

Knowing which layer each tool works at helps you avoid installing software to solve a process problem, or the other way around.

Key concepts

Understand → Migrate → Use both → Automate the discussion; claudex = software; Use Both = method; LGTM; complementary

What it is

The page gives a real example: someone with a Codex subscription uses its image generator, image_gen, which INEMA calls "image 2.5." Today at INEMA, Claude Code runs codex exec in non-interactive mode and asks image_gen for a PNG in a folder; if Codex fails or there is no credit, it falls back to the local model FLUX.2 klein. Each generation uses subscription credit, so it runs only when authorized. The image is checked afterward because text near the edges may be made up. The kit proposes using level 2 step by step: first confirm the tool and billing, never switch automatically to a paid API, save the actual file in the project, and open it at the size it will be used. In the page’s words, INEMA’s approach runs automatically in batches, with a local fallback; the kit’s approach is manual, with more cost checks and review.

Why learn it

Generated images have costs and visible errors. The two approaches show how to balance automation with review before using the file.

Key concepts

image_gen; codex exec; FLUX.2 klein fallback; check first, inspect afterward; save as a sibling version instead of overwriting

What it is

The page sums up how to start: "The best first step: a plan of your own, reviewed by the other one." Choose a small task, write the brief and acceptance criteria, ask one to make the plan and the other to critique it in the same file, for no more than two rounds, and create a handoff before closing the session. The area includes six courses: Claude → Codex (PT, EN, and ES), Codex Básico, Master Codex, iClaudeX, MakeClaudeX, and DeepClaudeX. It also includes kits and references: agente-claude-codex, Use Both, claudex on GitHub, Codex Cheat Sheet, and the topic map on INEMA.CLUB. The page cites an external video by Mark Kashef, which the Use Both kit follows, as a source of the author’s opinions, not a benchmark. His summary is in RESUMO-VIDEO.md in the kit repository.

Why learn it

A short first cycle shows the value of a second look in practice. The courses and kits show you where to go deeper at each step.

Key concepts

a plan of your own, reviewed by the other one; no more than two rounds; Codex Básico; Master Codex; iClaudeX; MakeClaudeX; DeepClaudeX; Use Both; claudex; Codex Cheat Sheet

View full version →

MODULE 2.3

OSWork: Which Edition Is for You?

Explain what OSWork is, what you build with it, and which of the five editions (v2, v6.2, v6, v5, Quick) fits each person’s experience and available time.

0% 0 of 0

What it is

OSWork presents itself as “AI as a work system” and “from chat to your agent environment.” An agent is an AI that carries out tasks using tools and files, rather than only replying in a conversation. The course teaches you to organize files, instructions, memory, and tools into a work system you can verify. The page’s central line is: “Your next conversation can start where the last one left off.” To do that, OSWork teaches you to turn loose requests into organized work: define the goal, choose the context, do a small task, and check the result. You start in chat and work your way up to a restricted bot and supervised operation on a VPS, a server rented online.

Why learn it

Without organization, every conversation with AI starts from scratch. A system with files and rules keeps track of what’s already been decided and makes the result verifiable.

Key concepts

work system; from chat to an agent environment; goal, context, small task, check; restricted bot; supervised VPS

What it is

The page offers “five ways to organize your work with AI” and asks you to choose based on your experience and available time. OSWork v2 is the technical path. OSWork v6.2 is the complete, step-by-step path. OSWork v6 is for beginners and takes a visual approach. OSWork v5 is for beginners 40+. OSWork Quick gets straight to the point. According to the page, courses and videos are available in Portuguese, Spanish, and English; through the area access, v6.2 and v6 have their own links in EN and ES.

Why learn it

Choosing the right edition can keep you from giving up at the first terminal or getting bored with the basics. Each path covers the same idea at a different level of depth.

Key concepts

v2 technical; v6.2 complete, step by step; v6 visual for beginners; v5 for beginners 40+; Quick gets straight to the point

What it is

OSWork v2 is the technical path: eight modules, four tracks, and 48 topics. It goes from chat to the terminal and Codex, files, Skills, Git, a Telegram bot, and supervised VPS operation. It also includes a practice kit. The eight modules are: Models; Chat, Work, and Desktop; Terminal and Codex in Practice; Folders, Markdown, and Secrets; AGENTS, Skills, and Memory; Git and GitHub Without Losing Work; Telegram as a Work Interface; and VPS from Scratch and 24/7 Operation. OSWork v6.2 is the same path redesigned for beginners: eight modules and 48 lessons of about 15 minutes each, with visual step-by-step guidance and a glossary of technical terms. In v2, students can mark topics as read, record questions, take notes, and export their learning journey; progress is saved in the browser and kept separate by language.

Why learn it

If you want to get as far as a bot and server, you need the full path. v6.2 makes that path accessible even without a technical background.

Key concepts

v2: 8 modules, 4 tracks, 48 topics, kit; v6.2: 48 lessons of ~15 min, glossary; Codex; Skills; Git; Telegram; VPS

What it is

Three editions are for beginners and don’t require installation or programming. OSWork v6 has seven lessons of about 15 minutes each, with simulated chat screens, real before-and-after examples, and visual step-by-step guidance. You create folders by topic, a sheet with your rules, a request template, and a review routine, without installing anything. OSWork v5 is for professionals 40+ who aren’t very familiar with AI: seven lessons of 20 to 25 minutes on organizing folders, guidelines, reusable requests, and a review routine, with no programming. OSWork Quick is the short, visual version for professionals 40+: seven lessons and about 77 minutes of study, covering the same topics as v5 with less text. Quick is also available as a video, with Nei’s avatar and voice, illustrations, and SRT captions in Portuguese, Spanish, and English.

Why learn it

For beginners, the value comes from organizing and checking your work, not from technical tools. These editions teach you to do that in a short time.

Key concepts

v6: 7 lessons of ~15 min, visual, no installation; v5: 7 lessons of 20 to 25 min, 40+; Quick: 7 lessons, ~77 min, also available as video

What it is

The page lists what you build along the way: a project folder with inputs, outputs, and its own instructions; a record of decisions and failures, along with a reusable Skill; a Git history for reviewing and recovering changes; and a query bot with sample data and an operations plan. A Skill is a packaged procedure an agent can reuse; Git is a system that keeps file versions. The kit section says: “Your first environment starts with real files.” The kit includes templates for AGENTS.md, memory, decisions, a failure log, a task contract, a report Skill, and a VPS plan. It also includes a Python bot with an offline self-test. The practice bot is deterministic: it doesn’t call an AI model or run messages as commands.

Why learn it

Finishing with concrete files, rather than just notes, makes what you learn reusable at work. The kit templates save you from starting from scratch.

Key concepts

project folder; record of decisions and failures; Skill; Git; query bot with sample data; ZIP kit; deterministic bot

What it is

The page advises: “Start with a small task.” There are three steps: choose the first module and compare two options available in your environment; create a practice folder and adapt the kit’s task contract; save the result, record a question, and return to your learning journey. In the “Before you start” section, the page explains that you can begin with the fundamentals without programming, and that the technical labs introduce the terminal, Git, and Python step by step. You can read the content without signing up; subscriptions, API usage, and VPS rentals are separate services that students choose for themselves. The course doesn’t promise an unsupervised autonomous agent: it teaches limits, verification, logs, and recovery.

Why learn it

A small task gives you a quick result and shows what still needs organizing. Knowing what costs extra helps you avoid surprises.

Key concepts

small task; practice folder; task contract; no programming at the start; read without signing up; paid services are separate; supervised operation

View full version →
Module complete