Workshops: Agentic Coding Tools in Depth

Advanced OpenAI Codex Workshop: Cloud Tasks, CLI & Code Review

Many teams have completed their first tasks with OpenAI Codex and then notice that results vary and nobody is sure when the CLI, the IDE or the cloud is the right choice. This workshop takes Codex further: reliable cloud environments, parallel tasks, well-structured AGENTS.md files, deliberate sandbox settings and automated code review on GitHub.

Who is this workshop for?

  • Development teams already using Codex who want to go beyond single prompts
  • Tech leads organising work across Codex cloud, CLI and IDE
  • Engineers responsible for repository setup and CI environments
  • Maintainers who want an automated first-pass review
  • Companies with ChatGPT Business or Enterprise workspaces rolling out Codex more widely

What you will take away

  • Configure cloud environments with setup scripts, dependencies and controlled internet access so tasks run reliably.
  • Delegate several cloud tasks in parallel, compare candidate solutions and turn the best one into a pull request.
  • Combine the CLI and IDE extension with cloud tasks, moving work deliberately between local and cloud.
  • Write layered AGENTS.md files that give Codex the build, test and style conventions of each part of the repository.
  • Choose approval and sandbox settings per situation and set up Codex reviews on GitHub pull requests.

Agenda

  1. Module 1

    Using the right Codex surface

    • Cloud tasks, CLI, IDE extension and code review compared
    • Local or cloud: speed, isolation, data and cost
    • Connecting the ChatGPT workspace to GitHub
  2. Module 2

    Environments for cloud tasks

    • Setup scripts, dependencies and secrets handling
    • Internet access for the agent: off, allowlisted or open
    • Exercise: run your repository's tests reliably in the cloud
  3. Module 3

    AGENTS.md and parallel delegation

    • Nested AGENTS.md files in monorepos
    • Writing tasks that work without follow-up questions
    • Exercise: delegate three backlog tickets, review them and merge as PRs
  4. Module 4

    CLI, sandbox and approvals

    • Approval and sandbox modes in the CLI and IDE
    • Configuration files, profiles and MCP servers
    • Non-interactive runs in scripts
  5. Module 5

    Code review with Codex

    • Requesting reviews in pull requests and enabling automatic reviews
    • Review guidelines in AGENTS.md
    • Combining Codex review with human review and CI checks

Typical friction after the first steps

Cloud tasks rarely fail because of the model. They fail because of the environment: missing dependencies, tests that expect local services, or conventions that are written down nowhere. We fix these points in your actual repository, so Codex works with your build and test setup instead of guessing.

Security settings you should understand

Codex runs cloud tasks in isolated containers and, locally, works in a sandbox whose strictness you choose. The defaults are sensible, but teams should decide consciously when the agent may access the network, write outside the workspace or run commands without asking.

  • No secrets in setup scripts or task prompts
  • Internet access only for required package registries and domains
  • Full-access modes only in disposable environments
  • Company code only in Business or Enterprise workspaces with admin controls

Codex alongside other agents

Many organisations use more than one agent. AGENTS.md is read by several tools, and we show how to keep it consistent with CLAUDE.md for Claude Code or Cursor rules. neooutput AI Code Engineers lead the day remotely or on-site, in English or German.

Frequently asked questions

Is this workshop suitable for Codex beginners?

It assumes experience with Codex or a comparable coding agent. Beginners are better served by our introductory OpenAI Codex workshop; on request we combine both.

Does our code leave the company with cloud tasks?

Cloud tasks run in OpenAI-hosted containers with a copy of your repository. Which data commitments apply depends on your ChatGPT plan and workspace settings. We explain the options; the GDPR assessment stays with your data protection team.

Can Codex replace our human code reviews?

No. Codex provides a valuable first pass that catches many issues early. Responsibility for what gets merged stays with your reviewers.

Workshop

Get in touch

Want Codex tasks to end in pull requests your team actually merges?

Tell us roughly how many people will join and pick dates that work – we will get back to you within 24 hours with a proposal for content, format and schedule.

Prefer to talk first?

15 minutes, free and without obligation – with our AI Trainers and AI Code Engineers.

Book a call for free (opens in new tab) +49 761 519 751 50 (24/7)

Request a training / workshop

Workshop: Advanced OpenAI Codex Workshop: Cloud Tasks, CLI & Code Review · 1 day

Dates that work for you (optional)

Suggestions about 2 weeks ahead – pick as many as you like

Reply usually within 24 hours · jocapps is ISO 27001 certified · Privacy