Workshops: Secure Agentic Coding & Governance
Privacy by Design with Coding Agents: GDPR Workshop for Developers
Coding agents read whatever is within reach: database dumps, log files, test fixtures, tickets containing customer emails. And they write code that happily logs full request objects or stores more fields than needed. This half-day workshop shows developers how to keep personal data out of agent context and how to steer agents toward code that implements data protection by design and by default, as Article 25 GDPR requires.
Who is this workshop for?
- Backend and full-stack developers handling personal data
- Tech leads of privacy-relevant products
- Data protection officers who want to understand agentic development
- QA engineers who build and maintain test data
- Product owners for applications in HR, health, finance or e-commerce
What you will take away
- Identify the paths through which personal data reaches an agent: files, logs, tickets, screenshots and MCP connections.
- Replace production data in development with synthetic or properly anonymised test data generated with agent support.
- Write project rules that make agents apply data minimisation, retention limits and pseudonymisation by default.
- Review agent-generated code for typical privacy flaws such as over-logging, unnecessary fields and unvetted third-party SDKs.
- Have agents draft data flow descriptions from code as input for your record of processing activities and DPIAs.
Agenda
Module 1
Personal data within the agent's reach
- Typical leaks: .env files, database dumps, logs, support tickets, screenshots
- Account types, data processing agreements and where the model runs
- Ignore files, permission rules and read-only MCP access
Module 2
Synthetic test data
- Exercise: replacing a production-like fixture with synthetic data
- Anonymisation versus pseudonymisation: what really removes the personal reference
- Seed scripts and factories that agents can extend safely
Module 3
Privacy rules for generated code
- Data minimisation, purpose limitation and deletion periods in CLAUDE.md or AGENTS.md
- Logging without personal data, field-level encryption, access control
- Exercise: reviewing an agent pull request through a privacy lens
Module 4
Documentation and working with the DPO
- Having data flows described from the code
- Input for the record of processing activities and the DPIA
- Privacy checks in the definition of done
What privacy by design means in an agentic workflow
The GDPR asks organisations to build data protection into processing from the start and to choose privacy-friendly defaults. With coding agents, that principle applies twice. First to the development process: the agent and its provider should only see the data they actually need. Second to the result: the code the agent writes should collect, store and log as little personal data as possible.
Agents learn the second part surprisingly well once you state your expectations. Clear project instructions, examples of good patterns and an automated check for personal data in logs change their output noticeably.
Data protection officers explicitly welcome
We translate the requirements into engineering practice; the legal assessment stays with your data protection officer and legal team. We recommend inviting your DPO to the final module, where documentation and approval processes are on the agenda.
The workshop is delivered by neooutput AI Code Engineers, remotely or on-site, in English or German, using sandbox repositories if your code must not leave the company.
Frequently asked questions
Can we use real customer data if we have a data processing agreement with our AI provider?
A data processing agreement is necessary, but it does not make every use proportionate. Data minimisation still applies: if synthetic data serves the development purpose, real personal data in prompts is hard to justify. We show practical ways to work without it.
Is anonymised data always safe to use?
Only if the personal reference is genuinely removed. Many supposedly anonymised dumps are in fact pseudonymised and therefore remain personal data under the GDPR. We cover simple tests that reveal this.
Don't agents write privacy-friendly code anyway?
Not reliably. They follow the patterns in your codebase and in common examples, which often log too much or collect extra fields. Project rules and review checklists make the difference.
Workshop
Get in touch
Want your coding agents to write privacy-friendly code without ever seeing customer data?
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: Privacy by Design with Coding Agents: GDPR Workshop for Developers · Half day (approx. 4 hours)
More workshops
- Workshops: Secure Agentic Coding & GovernanceEU Cyber Resilience Act for Developers: Agentic Coding Workshop
- Workshops: Secure Agentic Coding & GovernanceNIS2 Workshop for Dev Teams: Secure Development with AI Agents
- Workshops: Secure Agentic Coding & GovernanceMCP Server Governance: Vetting, Allowlists & Approval Workshop
- Workshops: Secure Agentic Coding & GovernanceWhen a Coding Agent Goes Wrong: Incident Response Workshop
- Workshops: Secure Agentic Coding & GovernanceTechnical Debt in AI-Generated Code: Maintainability Workshop
- Deploy & App StoresPush, In-App Purchases and Deep Links for AI-Built Native Apps
