Founders and operators
Use Codex to turn messy goals into plans, implementation tasks, review evidence and a clean handoff without pretending every task is a coding task.
Free OpenAI Codex tutorial
Make Codex
reliably useful
Give Codex the right operating context: checked-in rules, reviewed decisions and optional recall. Then verify each change before accepting it.
Required team rules belong in checked-in AGENTS.md and documentation. Local Codex memories are optional recall, not the source of a required rule.
Choose the right work surface
AGENTS.md, reviewed decisions and local recall
Proof gates and safe monitor rules
Get the full tutorial
A detailed, practical guide you can use while Codex is open beside your repo.
- Prompts for messy briefs, plans and verification
- AGENTS.md, reviewed decisions, local recall, skills and MCP
- Safe monitors and enterprise governance checkpoints
No spam. Just the tutorial and useful follow-ups.
Who this is for
A Codex workflow for the work around the code.
Developers and technical PMs
Learn the context hierarchy, isolated work, skills, MCP and subagent patterns that make a repo legible to an agent.
Teams and enterprise
Set a bounded pilot, clear permissions and human review before connecting shared company systems.
Inside the tutorial
The table of contents is a working system.
Turn messy intent into a contract
Start with the raw brief, then make the outcome, scope, evidence and approval boundary explicit.
Put each kind of context in its layer
Use checked-in AGENTS.md and docs for required rules, reviewed records for shared decisions, and local memories only as optional recall.
Add skills and MCP
Grow manual successes into skills and connect tools only where permissions, schemas and failure modes are understood.
Parallelize with control
Use plans, isolated worktrees and subagents for independent work, then consolidate with explicit checks.
Run goals and monitors safely
Use a verifier, material-update rule, stop condition and human approval instead of open-ended autonomy.
Ship evidence, not narration
Run type, unit, build, security and browser checks; report what changed, what passed and what remains uncertain.
The operating loop
From vague request to durable, trustworthy result.
Frame
Turn the raw brief into an outcome, scope, constraints and definition of done.
Context
Load the relevant docs, repo rules, project thread, tools and current system facts.
Work
Let Codex make the smallest coherent change in an isolated surface, with an explicit verifier.
Prove
Verify behavior, inspect the diff and leave a handoff another person can trust.
Codex memory versus AGENTS.md
Put required truth where the whole team can review it.
Codex needs different kinds of context for different jobs. Keep the boundary clear so a missing local memory never becomes a missing rule.
Required rules
Keep shared working agreements, commands and safety boundaries in checked-in AGENTS.md and nearby docs.
Reviewed state
Keep decisions, handoffs and active project facts in the system your team reviews and maintains.
Optional recall
Use local Codex memories as helpful recall. They are generated state, not the only place a required rule should live.
Teams and enterprise
The governance layer is part of the tutorial.
The guide includes a team path: start with a bounded pilot and make access, approval and evidence visible before connecting company data.
Scope the pilot
Choose approved users, repositories and task types. Name the owner and the exit criteria before expanding access.
Set access boundaries
Review the connected systems, read/write actions, approval points and least-privilege defaults with the responsible admin.
Review and rollout
Use a small pilot, approved repos, human review and evidence-based exit criteria. Never let a successful demo become an ungoverned production process.
Questions the tutorial answers
Is Codex only for software engineers?
No. The same framing and verification loop works for technical PMs, founders and operators working with a repo or a structured deliverable.
How is this different from a prompt list?
It teaches the surrounding system: checked-in rules, reviewed decisions, optional recall, permissions, skills, isolation, goals and proof.
Can I run a monitor continuously?
Only with a defined goal, verifier, cadence, material-update rule, stop condition and owner. It must not bypass approval boundaries or take external actions without permission.
What should I bring?
One real task, the relevant repo or documents, and a willingness to inspect the result instead of accepting confident prose.
Created by Iwo Szapar
Built from real agent workflows.
Iwo builds AI operating systems for knowledge work and maintains production workflows where context, permissions and verification matter as much as the model.
Give Codex a system to work inside.
Get the tutorial, choose one real task and build your first evidence-gated Codex loop.
Get the full tutorial