Workshops: Agentic Coding

Agentic Coding for Backend & API Development: Workshop

In the backend, data models, interface contracts and security determine quality, and that is exactly where AI agents need firm guardrails. This workshop shows backend teams how agents can implement APIs, migrations and integration tests reliably without breaking contracts or opening security holes.

Who is this workshop for?

  • Backend developers (Java, Kotlin, .NET, Node.js, Python, Go, PHP)
  • API and integration teams
  • Software architects
  • Teams running microservice landscapes
  • Platform teams for internal APIs

What you will take away

  • Have agents implement endpoints from OpenAPI or GraphQL specs while keeping the contract stable.
  • Run database migrations with agents safely, including rollback and data checks.
  • Use test-driven loops so the agent verifies its own changes.
  • Spot typical security flaws in agent-written backend code, such as authorisation and input validation.
  • Connect databases, logs and observability via MCP as context for the agent.

Agenda

  1. Module 1

    Backend context for agents

    • Architecture rules, layers and conventions in CLAUDE.md or AGENTS.md
    • Claude Code, OpenAI Codex, Cursor and GitHub Copilot in the backend
    • Local environment and test database for the agent
  2. Module 2

    API-first with agents

    • Exercise: endpoint from an OpenAPI spec including validation and error cases
    • Contract tests and breaking change detection
    • Keeping API docs current automatically
  3. Module 3

    Data and migrations

    • Schema changes with migration and rollback
    • Generating test data and edge cases
    • Query optimisation with EXPLAIN and agent support
  4. Module 4

    Test-driven agent loops

    • Unit, integration and contract tests as feedback
    • Testcontainers and reproducible environments
    • When the agent must not touch tests
  5. Module 5

    Security and operations

    • Checking authorisation, injection, secrets and logging
    • Observability via MCP: logs and traces as context
    • Review checklist for backend pull requests

Why backend teams benefit most

Backend code is easy to verify automatically: tests, types, schemas and contracts give the agent clear feedback. Provide these signals in a structured way and you can hand agents surprisingly large tasks.

At the same time, mistakes are costly. A faulty migration or a missing permission check can put data at risk. So we combine speed with clear approval points.

How this differs from general agentic coding workshops

This workshop is for teams who already know agents in principle and now want to master backend-specific patterns.

  • Focus on contracts, data and security
  • Exercises in your language and framework
  • Ready-to-use project rules and review checklists

Trainers

neooutput AI Code Engineers and AI Solutions Architects who use agentic backend development in client projects. Remote or on-site, in English or German.

Frequently asked questions

Does this work with Java or .NET enterprise code?

Yes. Strongly typed languages with good tooling give agents reliable feedback. We adapt the exercises to your stack.

Should the agent access our database?

We recommend access only to local or test databases and read-only connections for analysis. Production data does not belong in the agent's context.

How do we stop the agent from editing tests to make them pass?

Through clear rules in the project instructions, separate ownership of tests and reviews that specifically check test changes. We practise this in the workshop.

Workshop

Get in touch

Want agents to build your APIs without breaking contracts?

Tell us a little about your team – we will get back to you within 24 hours with a proposal for content, format and dates.

Prefer to talk?

We only use your details to handle this request. See our privacy policy.