Published tracks

Five tools, one assessment standard.

Every track ends the same way: a task-based assessment inside the real tool, reviewed by a human before a credential is issued. What differs is the syllabus underneath.

Claude Code

Agentic coding & repo-scale workflows

Covers running Claude Code against a real repository: scoping a task, reviewing its diffs, and knowing when to stop it and take over.

Levels: 3
Format: Cohort + async
Assessed: In-tool

Codex

OpenAI's coding agent in CI and local flows

Focuses on Codex inside CI pipelines and local terminals — the failure modes are different from a chat-based assistant, and so is the review habit it needs.

Levels: 2
Format: Cohort + async
Assessed: In-tool

GitHub Copilot

Pair-programming inside existing codebases

A shorter track for teams that already have GitHub Copilot switched on and want everyone using it the same, deliberate way.

Levels: 1
Format: Self-paced + live clinic
Assessed: In-tool

Gemini

Multimodal assistance across code, docs & data

Covers Gemini's multimodal range — reading a screenshot, a spreadsheet, and a codebase in the same task — which is where most teams under-use it.

Levels: 2
Format: Cohort + async
Assessed: In-tool

Cursor

AI-native editor workflows across a full codebase

Covers working in Cursor's Composer and Agent modes across a multi-file change — where the editor's own context choices matter as much as the model's.

Levels: 2
Format: Cohort + async
Assessed: In-tool

Which track for which role?

Role	Track	Why
Engineers running agents on production repositories	Claude Code	Task framing, diff review and permissions in a live repository
Teams on OpenAI with CI-heavy workflows	Codex	Failure triage without weakening tests
Engineers who live in the editor	Cursor	Context selection and delegation boundaries in the IDE
Teams with GitHub Copilot already switched on	GitHub Copilot	Accept, edit or reject: consistent judgment on suggestions
Teams working across code, docs and data in Google's stack	Gemini	Mixed-input tasks and source-checking
Final-year students and college batches	Claude Code Foundations	A lighter assessed task for a whole batch

Most teams start with Claude Code training for teams: a 2 to 4 week cohort on your own repositories that ends in a proctored task.

Looking for Microsoft 365 Copilot rather than GitHub Copilot? They're different products with separate credentials. Here's which Copilot certification you need.

Training students instead of employees?

There's a Foundations-level Claude Code programme for colleges and universities, with a lighter assessment so a whole batch can go through it.

See the student programme

Running a tool that isn't on this list?

We write the curriculum and the rubric around your internal or unreleased tooling directly.

See how custom tracks work

Not sure which track fits your team?

Send us the tool and the headcount — we'll tell you which published track applies, or whether this needs a custom one.

Talk to a program lead