Example · 3 days
Engineering with coding agents
Get real work out of coding agents without losing control: review, tests, refactoring, security and design that hold up.
For engineering teams adopting AI coding tools.
Day 1
AI-assisted software development
Quarter day · Software craft · Intermediate
- AI coding tools compared: Cursor, Claude Code and others.
- Plan mode: plan, review, then implement.
- Reviewing AI-generated code properly rather than "vibe coding".
- Reusable skills for repeated tasks.
- Agentic workflows: reviewing PRs, checking build pipelines, writing test plans and plain-English change summaries, and updating Jira tickets.
Coding agents in practice
Quarter day · Software craft · Intermediate
- Project instructions, context and conventions an agent follows.
- Tests and guardrails that keep agents honest.
- Running several agents in parallel without collisions.
- Lab: ship a feature with an agent, and review its work.
Automated testing strategy
Quarter day · Software craft · Intermediate
- Unit, integration and end-to-end tests, and the balance between them.
- Tests that describe behaviour, not implementation.
- Test doubles without testing the mocks.
- Lab: test a feature at the right levels.
Testing legacy code
Quarter day · Software craft · Intermediate
- Finding seams in code that was not built for tests.
- Characterisation tests.
- Breaking dependencies safely.
- Lab: put a harness around old code.
Day 2
Refactoring in small steps
Quarter day · Software craft · Intermediate
- Code smells and the refactorings that fix them.
- Small safe steps under test.
- IDE refactoring tools.
- Lab: refactor a real tangle in pairs.
Clean code and design
Quarter day · Software craft · Intermediate
- Naming, functions and boundaries.
- Coupling, cohesion and dependencies.
- Structure that follows the domain and the UI.
- Lab: review and improve a module together.
Secure coding: the OWASP Top 10
Quarter day · Software craft · Intermediate
- Injection, broken access control and the rest of the Top 10.
- How each shows up in your stack.
- Secure defaults in your frameworks.
- Lab: find and fix vulnerabilities in a sample app.
Secure coding: secrets, dependencies and threat modelling
Quarter day · Software craft · Intermediate
- Secrets management.
- Dependencies and supply-chain attacks.
- Threat modelling a feature in thirty minutes.
- Lab: threat-model a feature from your backlog.
Day 3
The LLM threat model
Quarter day · AI engineering · Advanced
- Why a model is neither a trusted component nor a simple input.
- Assets, trust boundaries and attackers for an LLM-backed system.
- The OWASP Top 10 for LLM applications.
- Exercise: threat-model a real feature.
Prompt injection and poisoned context
Quarter day · AI engineering · Advanced
- Direct and indirect prompt injection.
- Poisoned documents, retrieval and tool results.
- Mitigations that help, and the ones that only look as if they do.
- Lab: attack a RAG assistant, then defend it.
System design: requirements and building blocks
Quarter day · Software craft · Advanced
- Requirements, constraints and back-of-envelope sums.
- Caching, queues and databases.
- Consistency and failure.
- Exercise: estimate load for a real system.
System design: design and defend
Quarter day · Software craft · Advanced
- Writing a design document others can review.
- Trade-offs, and saying them out loud.
- Reviewing someone else’s design.
- Lab: design a service and defend it to the room.