Workshops: Secure Agentic Coding & Governance

Threat Modeling with and for Agentic Coding: STRIDE Workshop

Agentic coding changes threat modeling in two directions. Agents can derive data flow diagrams from code and prepare threat lists, so AppSec teams can cover far more systems. At the same time the agent is itself a new actor with rights, data access and outbound interfaces, and its setup needs a threat model of its own. This workshop covers both, using your real systems.

Who is this workshop for?

  • Application security teams and security architects
  • Security champions embedded in product teams
  • Software architects
  • Platform teams providing coding agents
  • ISMS and risk management owners

What you will take away

  • Have an agent derive a data flow diagram with trust boundaries from code and infrastructure definitions, and review it critically.
  • Produce an agent-assisted STRIDE analysis and separate solid threats from plausible-sounding noise.
  • Model your own agentic coding setup with its actors, assets and trust boundaries.
  • Derive prioritised countermeasures and link them to tickets and controls.
  • Keep threat models current as code in the repository instead of redrawing them once a year.

Agenda

  1. Module 1

    Threat modeling in brief

    • The four key questions: what are we working on, what can go wrong, what are we going to do about it, did we do a good enough job?
    • STRIDE, data flow diagrams and trust boundaries
    • Complementary methods: attack trees and LINDDUN for privacy threats
  2. Module 2

    Agents as threat modeling assistants

    • Exercise: derive data flows from code, API specifications and IaC
    • Having STRIDE-per-element and attack trees proposed
    • Spotting invented threats and blind spots
  3. Module 3

    Modeling the agent setup

    • Actors: developer, agent, model API, MCP servers, CI runners, cloud agents
    • Assets: source code, secrets, production access, customer data
    • Exercise: draw the trust boundaries of your own setup
  4. Module 4

    STRIDE applied to coding agents

    • Spoofing and tampering: agent identity, commits, manipulated rules files
    • Repudiation and information disclosure: missing logs, code at the provider, secrets in context
    • Denial of service and elevation of privilege: runaway costs, overly broad permissions
  5. Module 5

    Countermeasures and living models

    • Prioritising measures and accepting residual risk consciously
    • Threat modeling as code in the repository
    • Agent review for pull requests that touch trust boundaries

Two perspectives in one day

In the morning you use agents such as Claude Code or OpenAI Codex as a tool to build threat models for your applications faster and more completely. In the afternoon the agent becomes the subject of analysis: which paths does it open for attackers, and which controls are missing?

The two perspectives reinforce each other. Once you have watched an agent work, you model its risks more realistically.

Results you take home

You work on your own architectures, or on a sanitised example architecture for confidential systems.

  • A threat model of your agentic coding setup
  • A reusable template and prompts for agent-assisted threat modeling
  • A prioritised list of measures with owners
  • References to guidance such as the OWASP projects on LLM and agentic AI risks

Risk decisions stay with people

Agents provide suggestions, but they carry no accountability. Which threats matter and which residual risk is acceptable is decided by your team and your risk owners. This stance fits management systems such as ISO 27001, to which jocapps, the company behind neooutput, is certified.

Frequently asked questions

Do we need threat modeling experience?

No prior threat modeling practice is required if your team knows the basics of application security. We refresh the method briefly at the start so all participants share the same vocabulary.

Can an agent replace the AppSec review?

No. It speeds up the groundwork and catches things that get missed under time pressure. But it also invents threats and misses business context, so the assessment stays with experts.

Which tools do you use?

Alongside coding agents such as Claude Code or OpenAI Codex we work with diagrams as code, OWASP Threat Dragon or threat-modeling-as-code approaches. We build on whatever your team already uses.

Workshop

Get in touch

Does your agentic coding setup have a threat model of its own yet?

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: Threat Modeling with and for Agentic Coding: STRIDE Workshop · 1 day

Dates that work for you (optional)

Suggestions about 2 weeks ahead – pick as many as you like

Reply usually within 24 hours · jocapps is ISO 27001 certified · Privacy