Supply Chain Robots, Electric Sheep, and SLSA

A talk about creating automation, shifting left, attack vectors, attestations, verification, zero-trust, and SLSA.

In the talk I cover creating automation, shifting left, attack vectors, attestations, verification, zero-trust, and how the SLSA spec helps implement solutions for each. The main take away is that security needs to be applied everywhere in the pipeline. The talk should lead to a greater discussion around the challenges of securing the supply chain, supporting EO 14028 and ISO27001, and improving the security posture of your pipelines.

Attendee Takeaways

Answers for the following questions:

- Why do we need supply chain automation?
- What are common attack vectors in a supply chain?
- What techniques can we use to help secure the supply chain?
- What are the security benefits of supply chain automation and shift left?
- What specifications and tools can we use to help secure the supply chain?

https://jfrog.com/blog/swampup-session-highlights/#brett-smith

Event Provenance Registry: Continuous Delivery Events for the Electric Sheep

What if you got a second chance to build an Event Driven Provenance service? In this talk I will cover the decision to start over, rewrite, and Open Source the Event Driven system we built in house. In the process of covering the things we changed and the things we kept I tell a few war stories. Add in what needed to be improved and what we left behind. I will talk about our involvement in the CD Foundation and how the new system can leverage CDEvents and help with SBOM storage and retrieval. Demo and Discussion included dependent on time allotment.
Attendee Takeaways
Answers for the following questions:
- What does it take to open source in house tooling?
- What should I consider when open sourcing internal tooling?
- Why we made the choices we made to open source an internal project?
- What is an Event Driven Provenance Service?
- What is the CD Foundation?
- What is CDEvents?

30 minutes
The project https://github.com/sassoftware/event-provenance-registry

Workshop: Building an Event-Driven CI/CD Provenance System

In this hands-on workshop participants will journey through the architecture of an Event-Driven CI/CD Provenance System. We will not only cover microservice architectures, but also asynchronous communication, data interoperability, message specifications, and schema validation.

We will learn how to leverage Golang for service and CLI development, Docker for seamless deployment, Redpanda as a Kafka-compatible message bus, and PostgreSQL for efficient backend storage. The workshop uses the open-source project Event Provenance Registry (EPR) as the central service to leverage these technologies.

Over the course of the session we will delve into the EPR codebase, work through coding and building Golang services, discuss the theories of event driven systems, cover some pitfalls, and examine the integration with Redpanda for effective event propagation.

The workshop provides a valuable blend of theoretical understanding and hands-on experience in the dynamic landscape of Event-Driven CI/CD architectures.

4 90 minute sessions for the full workshop. The workshop can be modified to fit a smaller time slot.

First public delivery at DevOpsCon San Diego 2024

Secure the AI: Protect the Electric Sheep

In this session I go over how AI presents security risks to the Software Supply Chain, SDLC, developers, and architects. I cover attack vectors in the supply chain and how they relate to the OWASP Top 10 for LLMs as well as how they tie into scenarios in your CI/CD pipelines. We wrap up the session covering techniques to close the attack vectors and protect your pipelines, software, and customers.

Attendee Takeaways

Answers for the following questions:
- Why do we need to secure the AI?
- How do we secure the AI?
- What is the OWASP Top 10 for LLMs?
- What are the AI attack vectors in the supply chain?
- How do we close the AI attack vectors?

Session is 45 minutes.

Wrangling Third Party Dependencies: Are the Electric Sheep Healthy?

A talk about how we are working on curating our Third Party Dependencies using automation and online resources like Ecosyste.ms, deps.dev, OpenSSF Scorecard as well as Snyk, Sonatype, and others. What libraries are we using? What libraries are unsupported, abandoned, outdated, etc...? What open source tools can we leverage to help answer these questions and more?

Attendee Takeaways

Answers for the following questions:
- Why do we need to curate Third Party Dependencies?
- How to find libraries are we using?
- What libraries are unsupported, abandoned, outdated, etc...?
- What open source tools can we leverage to help answer these questions and more?

30 - 45 minute session. First presented at The 4th Annual North Carolina Cybersecurity Symposium 2025

Platform Engineering: Herding the Electric Sheep

A talk about platform engineering, DevOps, DevSecOps, sprawl, chaos, compliance, and security. Why engineer an Internal Developer Platform when I have DevOps? DevOps works fine when you are a 20 person start-up but it often doesn't scale to Enterprise level development efforts. When you have 3000 developers with different needs and you are responsible for EO compliance and security a modular self-service platform is a good choice to build. In this talk I cover the challenges we have faced in a 3000 developers enterprise and how we are working to address them. I also cover how we are working on automating, integration, and scaling the creation of our internal developer platform. We talk about the tools we are using and the good and bad decisions I have made along the way. I also talk about how we are leveraging SBOMs, SLSA, and other tools to help build out a secure and compliant platform. Attendees will learn the benefits and challenges of Platform Engineering

Attendee Takeaways

Answers for the following questions:
- Do we need a Platform Engineering Team?
- Is an IDP the right solution for my situation?
- What does a large scale IDP look like?
- What does it take to support a large scale IDP?
- What does security and compliance look like in an IDP?

30-45 minute session

Agents and MCP Servers: Are the electric sheep safe?

We have a new AI attack service. MCP servers are everywhere, and they are the new attack surface. Can the MCP server help protect the electric sheep from rogue agents and bad actors, or are they just another way to attack them? Agents are already being used to automate the software development lifecycle (SDLC), but they also introduce new risks. This talk explores the new attack surface created by MCP servers and agentic AI, focusing on potential vulnerabilities and mitigation strategies. We will discuss how agentic AI can enhance the SDLC while also addressing the security risks it introduces. The talk will cover the role of MCP servers in managing these risks and provide strategies for securing them against potential attacks.

Attendee Takeaways

Answers for the following questions:
- What does Agentic AI in the SDLC look like?
- What Security risk do Agentic AI bring to the SDLC?
- How can MCP servers help with Supply Chain Security?
- What are the risks of using MCP servers?
- What are strategies to mitigate attacks on MCP servers?

25 minute session

AI Accelerated Chaos: Are the Electric Sheep Doomed?

In the visionary landscape of a not so distant future, the "electric sheep" we
dream of are the revenue-generating products of our digital labor. To tend these
flocks, we have built sophisticated "supply chain robots"—agents of automation
within our CI/CD pipelines. But what happens when the shepherds themselves are
compromised? We are witnessing an unprecedented convergence of catastrophic
threat vectors accelerated by AI. This is not a theoretical drill anymore. It is
a perfect storm that is dismantling our security perimeters at machine speed.
We are witnessing an unprecedented convergence of three catastrophic threat
vectors accelerating security chaos. First, users downloading unverified AI
personal assistants that gain full system access, bypassing security controls
for convenience, unaware of the danger. Second, "Harvest now, decrypt later" is
no longer speculation retroactively compromising everything we protect now.
Finally, the Geopolitical AI Arms Race sees nation-states deploying vulnerable
models at scale, where high adoption rates.
The malware industry is adapting faster than our defenses. This session delivers
the uncomfortable truth: our supply chain robots and the electric sheep they
produce are at risk of being compromised by users who are unaware of the risks.
Join this session to understand this perfect storm and leave with a battle plan
to survive the next phase of cybersecurity chaos.
Attendee Takeaways
- Discover how local-first AI assistants are creating new attack surfaces.
- Learn how the Dunning-Kruger effect is creating a "security gap".
- Understand why "Harvest Now, Decrypt Later" makes PQC an immediate
infrastructure crisis.
- Gain information on the hidden security costs of viral foreign AI models.
- Leave with a concrete action plan covering immediate tactical fixes and
long-term strategic changes.

Cognitive Disarmament: When AI Coding Tools Erode Developer Skills

As we race to integrate AI coding assistants into every stage of the Software
Development Lifecycle, we are inadvertently creating a new, invisible security
vulnerability: the erosion of human intelligence itself. I call this
Cognitive Disarmament, the systematic atrophy of critical thinking,
debugging, and architectural judgment caused by an over-reliance on the
"prompt-and-accept" reflex.

While organizations obsess over supply chain attacks and zero-day exploits, a
more insidious threat is growing within their engineering teams. Developers are
becoming "middle managers" of code they do not deepy understand, skipping the
"productive struggle" essential for building expertise. This session explores
the dangerous convergence of decision-time compression, the metacognitive blind
spots created by AI confidence, and the loss of "deep work" capabilities. When
the AI goes down—or worse, when it hallucinates a subtle security flaw—will your
team have the mental fortitude to detect and fix it?

This keynote moves beyond the hype of AI productivity to address the existential
risk of skill degradation. We will discuss the concept of "Cognitive
Complementarity," a strategic framework where AI augments rather than replaces
human capability, and outline a battle plan to rebuild the mental resilience of
your engineering workforce before the next crisis hits.

Attendee Takeaways

1. The Hidden Supply Chain Risk: Understand how the "Google Effect 2.0"
transforms developers from creators to passive approvers, directly reducing
code comprehension and creating a new attack surface for subtle supply chain
vulnerabilities.

2. Combating the "Prompt-and-Accept" Reflex: Learn "active engagement"
techniques to transition developers from blind acceptance to critical review,
ensuring they remain the authoritative human-in-the-loop for
security-critical decisions.

3. Building Cognitive Resilience: Discover practical, controversial
implementations like "No-AI Fridays," "Mind Gyms," and "AI-Free Onboarding"
designed to maintain fundamental debugging skills and problem-solving muscle
memory.

4. Redefining Code Review: Transform code review from a compliance checkbox
into a high-friction learning opportunity that verifies deep understanding
("why it works") rather than just syntax correctness.

5. **The 30/70 Complementarity Framework** Apply a structural rule for dividing
labor: AI handles the routine 70% (boilerplate, tests), while you rigorously
protect the critical 30% (architecture, ethics, security design) that
requires human judgment.

Reproducible Builds: Robots recreate Electric Sheep

A talk about the security benefits and challenges of reproducible builds. It includes a real world comparison of the Debian and Fedora build systems and a discussion on the value based on the effort. Add in the work Fedora has been doing to increase reproducibility. Listeners should come away with knowledge of what reproducible builds are and opinions on if they are worth the effort.

Attendee Takeaways

Answers for the following questions:
- What is a reproducible build?
- Why do we need reproducible builds?
- What are the security benefits of reproducible builds?
- What are the security challenges of reproducible builds?
- What is the value of reproducible builds?

30 minutes plus QA discussion. First presented at the NCSU Secure Software Supply Chain Community Day, Presented at DevOps Con 2024 San Diego.