When does class start/end?
Class hours may vary, please reach out to contact@ascendientlearning.com if you have any questions.
DevSecOps for AI Coding Agents
Harden DevSecOps for AI coding agents across every lifecycle phase. AI coding agents are framed as an insider-threat surface, with the CIA triad applied to AI-generated code and the OWASP Top 10 for...
Read More
Duration
3 days
Course Code
AIC-2105
Available Formats
Classroom
Overview
Course Description
Harden DevSecOps for AI coding agents across every lifecycle phase. AI coding agents are framed as an insider-threat surface, with the CIA triad applied to AI-generated code and the OWASP Top 10 for Agentic Applications grounding the rest of the course. Plan, Code, and Build cover threat modeling with STRIDE and LINDDUN, AGENTS.md and CLAUDE.md as context contracts, licensing contamination, spec-driven design as an invariant-enforcement strategy, linting and type systems as agent guardrails, slopsquatting, and AIBOM hygiene. Test, Release, and Deploy cover mutation testing of weak agent assertions, rules-file-backdoor attacks, sandboxed agent execution, dual-LLM and capability-based runtime defenses, and EU-AI-Act-compliant audit trails. Operate, Monitor, and the human layer cover runtime prompt-injection defense, automation bias, agentic misalignment, risk-tiered review rubrics, and quality metrics that resist velocity theater. Hands-on labs produce CIA diagnostics, AGENTS.md hardening, mutation tests, rules-file-backdoor catches, sandbox-escape defenses, and a CI attribution check that yields an EU-AI-Act-compliant audit trail. The course is designed for software developers, technical leads, and security professionals deploying AI coding agents in production codebases.
Skills Gained
By the end of this course, participants will be able to:
- Map how AI coding agents pressure each phase of the DevSecOps lifecycle
- Configure spec-driven development, type systems, and schema validation as agent guardrails
- Design security testing that verifies agent output rather than confirms it
- Harden CI/CD pipelines and runtime against agentic AI’s unique threat vectors
- Establish attribution, provenance, and EU-AI-Act-compliant audit trails for AI-authored changes
- Restructure team workflows and metrics to resist rubber-stamping and velocity theater
Who Can Benefit
This course is designed for:
- Software Developers
- Data Scientists & Analysts
- Technical Leads
- Security Professionals
Prerequisites
Participants should enter this course with:
- Practical Python programming experience
- Completion of an AIC-110X course or equivalent experience with an AI coding agent
Organizational Objectives
This course assists organizations to:
- Reduce security defects in AI-authored code by enforcing types, schemas, and pre-commit gates
- Lower supply-chain risk through SBOM/AIBOM hygiene and slopsquatting detection
- Establish auditable agent-attribution trails that satisfy EU AI Act Articles 19 and 26
- Contain incidents from agent misbehaviour through sandboxing and runtime tool-use monitoring
- Build team-wide capability to review AI-generated code without rubber-stamping
Software
All attendees must have a modern web browser and an Internet connection.
Course Details
Course Details
Module 1 - AI Agents and the DevSecOps Lifecycle
By the end of this module, you will be able to characterize AI coding agents as an insider-threat surface, apply the CIA triad to AI-generated code, and locate where each DevSecOps phase needs existing practices versus emerging ones to handle agent-driven work.
- AI coding agents — an insider threat
- The CIA triad applied to AI-generated code
- OWASP Top 10 for Agentic Applications as grounding taxonomy
- How agents pressure each phase of the lifecycle
- Existing practices, what they prevent, emerging practices
- The team and culture thread running through every phase
- Hands-on Lab: Run a CIA-triad diagnostic on a recent AI-generated code change to identify which assurances hold and which don’t.
Module 2 - Plan: Threat Modeling and Secure Design
By the end of this module, you will be able to apply STRIDE and LINDDUN to AI-augmented systems, capture security and licensing requirements in architectural decision records, codify trust assumptions in AGENTS.md and CLAUDE.md context contracts, and use spec-driven development to enforce invariants.
- STRIDE and LINDDUN for AI-augmented systems
- Security requirements, ADRs, and licensing/copyleft contamination in AI-generated code
- When agents skip design and hallucinate mitigations
- Spec-driven development and design-by-contract
- AGENTS.md and CLAUDE.md: context contracts and their trust assumptions
- Invariant enforcement and formal verification
- Hands-on Lab: Author an AGENTS.md that hardens agent output by codifying trust assumptions, licensing rules, and design invariants.
Module 3 - Code: Secure Coding Under AI Acceleration
By the end of this module, you will be able to use linting, pre-commit hooks, type systems, and schema validation as deterministic guardrails on AI-generated code, organize the codebase to limit agent blast radius, and treat the IDE itself as a security surface.
- Linting and pre-commit hooks as enforcement that agents cannot bypass
- Type systems as agent guardrails
- Schema validation at trust boundaries
- Codebase organization to limit agent blast radius
- Secrets management with vaults and short-lived tokens
- The IDE as attack surface: browser runtime, extensions, and config files
- Hands-on Lab: Harden a small AI-generated code change by adding types, schemas, secret-rotation hooks, and pre-commit gates that the agent cannot bypass.
Module 4 - Build: Supply Chain Integrity
By the end of this module, you will be able to maintain SBOMs and AIBOMs, configure dependency review and allowlists, detect hallucinated package suggestions (“slopsquatting”), and keep lockfiles deterministic when agents edit dependencies.
- SBOM, dependency pinning, signed artifacts, and AIBOM (AI bill of materials)
- Dependency review and allowlists
- Slopsquatting and hallucinated package detection
- Verifying agent-suggested dependencies
- Lockfile hygiene and non-determinism in agent edits
- Case study: when your agent vendor is your supply chain
- Hands-on Lab: Run a slopsquatting hunt on a real requirements file and trace the agent prompts that introduced suspicious entries.
Module 5 - Test: Security Testing and Validation
By the end of this module, you will be able to wire SAST, DAST, and TDD into AI-assisted workflows; expose weak agent assertions through mutation and property-based testing; and protect existing tests from silent agent modification across refactors.
- SAST, DAST, and coverage metrics
- TDD as an AI guardrail
- Mutation testing to expose weak AI assertions
- Property-based testing for edge case discovery
- Approval testing and protecting existing tests from agent modification
- Iterative vulnerability drift across AI refactoring
- Hands-on Lab: Run mutation testing on an AI-generated code path and document the weak assertions it surfaces.
Module 6 - Release: Review and Approval Gates
By the end of this module, you will be able to enforce mandatory review and signed commits on AI-authored changes, configure AI reviewers with project-specific rules, use agent hooks as deterministic review gates, and recognize rules-file backdoor attacks against context files.
- Mandatory review, branch protection, and signed commits
- Rules file backdoors and hidden Unicode in context files
- Configuring AI reviewers with project-specific rules and multi-pass pipelines
- Agent hooks as deterministic review gates
- Testing and validating the review system against known attacks
- Defining the boundary between AI and human review
- Hands-on Lab: Catch a rules-file backdoor with an AI reviewer that runs against a poisoned context file.
Module 7 - Deploy: Change Management and Sandboxed Execution
By the end of this module, you will be able to integrate AI coding agents into change-management gates, validate IaC and policy-as-code, sandbox agent execution at process, container, and MicroVM levels, and use short-lived non-human identities for agent activity.
- Change advisory boards, approval gates, and IaC/policy-as-code validation
- Sandboxing agents — process, container, MicroVM, and cloud approaches
- CI/CD gates that agents cannot game
- MCP server authentication, sandboxing, and permission scoping
- Non-human identity and short-lived scoped tokens for agents
- Autonomous sandbox escape and unauthorized production changes
- Hands-on Lab: Sandbox an AI coding agent with a deliberately ambiguous task and defend the sandbox against escape attempts.
Module 8 - Operate: Runtime Defense Against Agent Threats
By the end of this module, you will be able to recognize prompt injection as an injection class, apply dual-LLM and capability-based defenses, harden MCP servers and isolate agent context, and monitor agent tool use at runtime to detect exfiltration.
- Prompt injection as a new class of injection
- Dual-LLM and capability-based defenses (CaMeL pattern)
- MCP server hardening and permission scoping
- Data classification, context isolation, and agent blast radius
- Hooks for runtime policy enforcement and tool-use blocklists
- Runtime monitoring of agent tool use and exfiltration detection
- Hands-on Lab: Defend a Claude Code agent against a prompt-injection attack delivered through a malicious GitHub Issue.
Module 9 - Monitor: Audit, Attribution, and Incident Response
By the end of this module, you will be able to instrument SIEM-grade audit logs for agent activity, satisfy EU AI Act Articles 19 and 26 log-retention requirements, attribute commits to human versus agent authors, and recover from incidents where the agent misrepresents what it did.
- Audit logs, SIEM, and anomaly detection
- EU AI Act Articles 19 and 26: log retention and audit trail requirements
- Attributing commits to human vs. agent authors
- SLSA provenance for AI-generated changes
- Tamper-evident logs for agent activity
- Incident detection, containment, forensics, and rollback for agent-caused failures
- Hands-on Lab: Build a CI attribution check that produces an EU-AI-Act-compliant audit trail for agent-authored commits.
Module 10 - Team: The Human Layer
By the end of this module, you will be able to recognize rubber-stamping, automation bias, and velocity theater inside an AI-heavy team; design risk-tiered review rubrics; and choose quality metrics that don’t reward speed at the expense of correctness.
- Rubber-stamping, automation bias, and velocity theater
- Agentic misalignment — insider-threat behaviour in frontier models
- Risk-tiered review: what to review every line, what to review by exception
- Team ownership and design authority when the agent wrote the code
- Quality metrics over velocity metrics
- Building shared understanding across agent-generated codebases
- Hands-on Lab: Build a risk-tiered review rubric and apply it to a mixed PR queue containing both human- and agent-authored changes.
Schedule
FAQ
Does the course schedule include a Lunchbreak?
Classes typically include a 1-hour lunch break around midday. However, the exact break times and duration can vary depending on the specific class. Your instructor will provide detailed information at the start of the course.
What languages are used to deliver training?
Most courses are conducted in English, unless otherwise specified. Some courses will have the word "FRENCH" marked in red beside the scheduled date(s) indicating the language of instruction.
What does GTR stand for?
GTR stands for Guaranteed to Run; if you see a course with this status, it means this event is confirmed to run. View our GTR page to see our full list of Guaranteed to Run courses.
How do I find an Ascendient Learning training location?
We have training locations across the United States and Canada - View a complete list of classroom training locations.
Which delivery formats are available?
At Ascendient Learning, we offer training that is Instructor-Led, Online, Virtual, and Self-Paced.
Does Ascendient Learning deliver group training?
Yes, we provide training for groups, individuals and private on sites. View our group training page for more information.
What does vendor-authorized training mean?
As a vendor-authorized training partner, we offer a curriculum that our partners have vetted. We use the same course materials and facilitate the same labs as our vendor-delivered training. These courses are considered the gold standard and, as such, are priced accordingly.
Is the training too basic, or will you go deep into technology?
It depends on your requirements, your role in your company, and your depth of knowledge. The good news about many of our learning paths, you can start from the fundamentals to highly specialized training.
How up-to-date are your courses and support materials?
We continuously work with our vendors to evaluate and refresh course material to reflect the latest training courses and best practices.
Are your instructors seasoned trainers who have deep knowledge of the training topic?
Ascendient Learning instructors have an average of 27 years of practical IT experience and have also served as consultants for an average of 15 years. To stay current, instructors spend at least 25 percent of their time learning new, emerging technologies and courses.
Do you provide hands-on training and exercises in an actual lab environment?
Lab access is dependent on the vendor and the type of training you sign up for. However, many of our top vendors will provide lab access to students to test and practice. The course description will specify lab access.
Will you customize the training for our company’s specific needs and goals?
We will work with you to identify training needs and areas of growth. We offer a variety of training methods, such as private group training, on-site of your choice, and virtually. We provide courses and certifications that are aligned with your business goals.
How do I get started with certification?
Getting started on a certification pathway depends on your goals and the vendor you choose to get certified in. Many vendors offer entry-level IT certification to advanced IT certification that can boost your career. To get access to certification vouchers and discounts, please contact info@ascendientlearning.com.
Will I get access to content after I complete a course?
You will get access to the PDF of course books and guides, but access to the recording and slides will depend on the vendor and type of training you receive.
How do I request a W9 for Ascendient Learning?
View our filing status and how to request a W9.
Reviews
It was very informative and covered all the required materials along with handson labs for practice.
Good training material and good instruction. More time needs to be provided for the lab work.
Sean is the very good instructor. I would like to take his class again in the future.
Great training it covered the most importan topics if GitHub copilot with good explanation and good labs.
great class and packed with material, would have lived to spread it more into many days but overall very informative.
