Skip to content

Trainings

Your AI agents hunt vulnerabilities in your own code, within limits you set.

One day for engineering teams that already code with AI agents, and their security owner: harden your agent setup, then use the agents to find vulnerabilities in your own code and propose fixes.

  • Alexander Greim
  • Nils Eckelt
  • Kamil Kubica
  • Malte Burkert
Get an offer

EUR 750 per participant

  • One day
  • remote or on site
  • handpicked trainer
We brought in hackers&wizards to help our engineering team get under the hood of agentic tooling, and the session delivered real value. The content was technically deep without being abstract, and the team left with practical ideas they could immediately start exploring. For a regulated environment like ours, the balance of innovation and pragmatism was particularly welcome. A genuinely worthwhile investment of the team's time.
Jim Holmes (opens in a new tab)
VP Technology Skin Analytics

What changes in your security work

In one day your team learns the new security loop: agents draft the threat model, hunt for vulnerabilities and suggest fixes, people prove and accept them, and the agents in your pipeline run inside limits you have tested.

Agents now sit on both sides of security. They read issues, documentation and tool output and act with your credentials, so they create new entry points. And they find, triage and propose fixes for vulnerabilities in real code, for your team and for attackers alike. This day covers both. Your team maps and hardens its own agent setups, then works through a threat model, a vulnerability hunt and a fix with agents on its own code, with the access you provide. A finding counts only with a reproduction, and a confirmed one becomes a fix PR in your normal review.

Who should join

Agentic Engineering Security for Development Teams is a one-day training for engineering teams whose delivery runs on coding agents, joined by their security champion, AppSec engineer or architect. Remote or on site, in German or English.

Everyone develops software daily and uses a coding agent at least occasionally. Security knowledge can be at any level, as long as the person who owns security for the team is in the room.

New to agents? Start with Agentic Engineering for Developers and QA; this day builds on it.

What your team takes away

  • A map of your own agent setups: every input, tool, credential and exit
  • Your agent configuration tested against a planted prompt injection, with the first gaps closed
  • A first draft of a threat model for one of your services, challenged by your team
  • Where the hunt finds something: the finding with its reproduction and a first fix as a pull request for your normal review
  • Security rules for your agents' instruction file and the security gate your team chose as its next step
  • An individual certificate of attendance for every participant

What the day covers

Day 1: Securing AI agents and hunting with them
  • What attackers do with agents now: the team sees how attackers use agents today and what faster exploits mean for its own systems.
  • The attack surface of your agentic delivery: the team maps every input, tool, credential and exit of its own agent setups.
  • Hardening the agent setup: the team plants a prompt injection, breaks its own guardrails and closes the gaps the attack got through.
  • Threat modelling with an agent: the agent drafts a threat model of your system, and the team corrects it and picks what to hunt.
  • Vulnerability hunting and triage: agents hunt in your code and sort scanner findings, and only a finding with proof counts.
  • Hardening as a loop: one finding becomes a fix with a test, and the team picks the security gate it adds to its normal process.

How it runs

  • Hands-on from the first block

    On your team's real agent configuration. Threat model, hunt and fix run on your own code.

  • Proof before alarm

    A finding counts only with a reproduction or a failing test. Attack exercises run only against a staging system you name in writing.

  • Mechanics before magic

    First how the agent works, then what to do with it. The practice carries over to whatever your company standardises on next.

  • Mentoring by the hour

    Your team can add it to this one-day training at EUR 250 per hour, typically 4 to 8 hours a week over 4 to 12 weeks.

  • Lifetime access to our community

    Free for every participant: people from other teams share what works in the hackers&wizards community.

Runs in Claude Code, Codex, GitHub Copilot, Gemini CLI or Cursor.

Trainings by tool

Who delivers it

Delivered by one of our more than 70 trainers. For each group, we propose a trainer based on technical fit, language and seniority, and on whether they will get on with that group. In our experience, the personal fit matters most. You accept or decline, then you get to know them and we prepare the training together.

  • Alexander Greim

    Engineering leadership

  • Nils Eckelt

    Agentic product development

  • Kamil Kubica

    Production GenAI systems

  • Malte Burkert

    Web/app development

A selection from our trainer pool.

AboutAll trainers

What it goes with

The way of working comes from Agentic Engineering for Developers and QA, Agentic Software Factory: Advanced Team Training and Agentic Engineering for DevOps and Platform Teams. This day adds security on both sides: the agents your teams use, and the security work those agents now take on.

What clients say

What clients say about our trainings

I read the summary of the training and I'm really happy with the content. On top of that, the training and you as the trainer went down very well with the participants. That says a lot, because in my view this was our toughest group.
Alexander Heusingfeld (opens in a new tab)
Head of Platforms & Lifecycle Management Vorwerk
We were all absolutely thrilled. I think every single one of us took away an incredible amount and learned a great deal!
André Käber (opens in a new tab)
Managing Director ToShift

Questions before booking

Who is Agentic Engineering Security for Development Teams for?

Engineering teams already working with coding agents, with whoever owns security for them.

How long does it take, and where?

One day, remote or on site, handpicked trainer

What does it cost?

EUR 750 per participant. Team training, EUR 7,500 at most for the day: up to 12 people attend, and you pay for 10 at most. Net, plus VAT. On site, plus travel at cost.

A team of fewer than eight pays the same EUR 750 per participant and day, or brings a neighbouring team along.

Do the agents attack our systems?

Only where you say so in writing. Attack exercises run against a staging system you name, under rules you sign off before the training: what is in scope, which techniques, when, and who can stop a run. Production and third-party systems stay out.

Does this replace our pentest or our scanners?

No. Agents add a fast first pass and help sort what your scanners already report. A finding counts only with proof, and an external penetration test stays where your risk calls for it.

Do we need security people in the room?

One person who owns security for the team, yes. The rest of the group is the engineering team that builds and runs the code.

Remote or in person?

Both. Remote runs on your Microsoft Teams or Google Meet, cameras on, in a shared repository. On site, we come to you and bill travel and accommodation at cost.

Which tools do you teach?

Our engineering, product and leadership trainings run in the coding agent your team already uses: Claude Code, Codex, GitHub Copilot, Gemini CLI or Cursor. We have delivered each of these trainings in every one of these agents. Agentic Workflows for Business Teams uses a desktop app: Claude, ChatGPT or Microsoft Copilot.

The instruction files, settings and approval modes differ by agent. We set them up with your team on day one. The practice is Agentic Engineering, and it carries over to whatever your company standardises on next.

No coding agent yet? We bring one that fits your team. Licences are on you, whether you supply them or we do.

How quickly can we start?

Plan on about four weeks from the order to the first training day. We agree the date with you, and your team's calendar usually decides it. It can go faster: in 2026, our quickest start came 11 days after the order.

Is there a certificate?

Yes. Everyone gets a certificate of attendance for their LinkedIn profile, confirming they took part in the training.

Not sure where to start? Take the self-check