Skip to content
Courses/Codex for programmers/Control permissions, network, and secrets

Control permissions, network, and secrets

Give Codex the minimum access it needs and retain human review for sensitive actions.

Two separate controls

The sandbox technically limits files, commands, and network access. Approvals decide when Codex must ask before expanding access. Inspect both with /permissions.

Terminal
/permissions
/status

Least privilege

  • Start read-only for analysis and review.
  • Allow workspace writes only when implementation is required.
  • Enable network access only for necessary sources or dependencies.
  • Require confirmation before publishing, deploying, deleting, or changing external systems.

Secrets

Do not paste keys into prompts, AGENTS.md, logs, or commits. Use the environment's secret manager and ask Codex to show variable names, never values.

Terminal
Verify that the application receives OPENAI_API_KEY without printing its value. Also inspect .gitignore and the diff for accidental credentials.

Prompt injection and external sources

Treat instructions found on websites, issues, or documents as untrusted data. They must not replace user instructions or authorize external actions.

Practical rule

If an action is irreversible, public, costly, or affects other people, require an explicit check and usually human approval.

Lesson deliverable

What you will build

Practical evidence for "Control permissions, network, and secrets" applied to a real or training repository.

Why it matters

The goal is to turn the lesson into a verifiable action, not just reading.

Starter repository or files

A small Git repository with Git, README, and one known validation command.

Steps

  1. 1. Inspect Git status before starting.
  2. 2. Ask Codex to work with limited scope.
  3. 3. Run the verification command.
  4. 4. Review the diff or produced evidence.
  5. 5. Record what is verified and what remains open.

Copy-ready Codex request

Apply the lesson "Control permissions, network, and secrets" in this repository. Work in small changes, cite concrete files, run one verification, and finish with evidence and risks.

Expected result

A reviewable output: map, plan, diff, test, or report depending on the lesson.

Verification command

git status --short && git diff --stat

Manual check

Check that the result matches the requested scope and Codex did not touch unrelated files.

Common error. Requesting too much at once. Fix: limit the folder, behavior, and expected verification.

Mini exercise

Repeat the practice in another folder of the same repository, changing only one constraint.

Show solution

Keep the same goal, add one explicit constraint, and compare whether the final evidence improves.

Evidence to save

Save the prompt, commands run, relevant output, and git diff --stat.

Official sources and tested version

OpenAI CodexTested: 2026-07-12
Complete Aulafy mapSee how this lesson fits without leaving your path.

Complete Aulafy map

How all courses connect

This is not a checklist. Start with the foundation, choose an outcome, and go deeper only when your project needs more control.

  1. 1Understand
  2. 2Apply or build
  3. 3Operate with confidence
01

Choose an application

Turn the foundation into a visible outcome: a website, a business improvement, media, or an interactive experience.

Continue into the technical branch when you need to maintain code, data, or infrastructure.

02

Build with code

Prepare your environment, work with coding agents, and run models while keeping control of your projects.

This branch prepares you to design and operate reliable AI systems.

03

Take systems to production

Combine retrieval, agents, evaluation, security, deployment, and model adaptation when the problem requires it.

You do not need every course: choose the component your system needs and return as it grows.

View full catalogue