From Alert to Fix: How AI Agents Are Changing Dependency Remediation

When the Axios npm package was poisoned in March 2026 reaching 100 million downloads per week teams with automated dependency remediation closed their exposure window in hours. Teams without it spent days manually triaging alerts.

Dependency vulnerabilities don't wait for human review cycles. But traditional security tooling still treats remediation as a manual, linear process: alert fires → engineer reads it → engineer opens a PR → engineer merges it. That loop is too slow when attacks cascade across five package ecosystems in two weeks, as the TeamPCP campaign demonstrated this year.
This session breaks down how AI agents are collapsing that loop. Drawing on the architecture of AI-assisted dependency remediation at scale across 20+ package ecosystems and hundreds of millions of repositories, I'll cover:

• How AI agents triage dependency alerts, separating exploitable vulnerabilities from noise

• How agents handle the hard cases: breaking changes, transitive dependencies, and packages with no safe upstream patch

• The attack surface AI introduces when remediation automation itself becomes a target, as prt-scan's 500+ malicious PRs against GitHub repositories demonstrated

• What "autonomous remediation" actually looks like in production: the guardrails, failure modes, and human-in-the-loop design decisions that matter

Attendees leave with a concrete mental model for where AI accelerates dependency security, where it still needs a human, and how to build pipelines that don't trade speed for correctness.

The Invisible Weak Link: Inside Modern Software Supply Chain Attacks

Last week, LiteLLM a Python library downloaded 95 million times a month was silently compromised through an attack that first weaponized a security scanner. In 2021, Log4j sent every tech company into emergency mode. In 2020, SolarWinds exposed 18,000 organizations including multiple government agencies. These are not isolated incidents. They are symptoms of a systemic crisis: the software that runs modern life is assembled from thousands of open-source ingredients, and attackers have learned that poisoning the supply is far more efficient than breaking through the front door.

In this session, you will learn how modern supply chain attacks actually work from dependency confusion and typo-squatting to CI/CD pipeline compromise and credential cascading. Using real-world case studies from 2024–2026, we will dissect the anatomy of these attacks, understand why they are getting worse (deeper dependency trees, AI coding assistants recommending popular but unvetted packages, and an explosion in open-source consumption), and explore the automated defense strategies that organizations can deploy today. Attendees will leave with a concrete framework for assessing their own supply chain exposure and actionable steps to reduce it.

Format: Session Presentation (30 min) or Keynote (45 min)
Level: Introductory to Intermediate

Securing 180 Million Developers: Lessons from Operating Dependabot at Scale

Dependabot is the most widely adopted dependency management tool in the world. It monitors over 7 million repositories across 20+ package ecosystems, opens more automated pull requests than any other user on GitHub, and draws from the GitHub Advisory Database with over 28,000 reviewed security advisories. But operating a security tool at this scale introduces challenges that most teams never encounter: how do you reduce false positives across dozens of language ecosystems? How do you prioritize millions of alerts so developers act on what matters? How do you handle the cascading complexity of transitive dependencies?

In this talk, I will share hard-won lessons from leading the Dependabot team at GitHub what works, what breaks, and what keeps us up at night. We will cover how we approach alert fatigue reduction, compatibility scoring, grouped security updates, and the emerging role of AI in vulnerability triage and remediation. Whether you are a security engineer trying to scale your own dependency management programme, an engineering leader evaluating SCA tools, or an open-source maintainer navigating the advisory ecosystem, this session will provide practical insights you can apply immediately.

Format: Session Presentation (30 min)
Level: Intermediate to Advanced

AI Meets AppSec: The Next Frontier in Automated Vulnerability Remediation

The average application today contains 581 open-source vulnerabilities a 107% increase in just one year. AI coding assistants recommend popular libraries without evaluating whether those libraries are secure. Meanwhile, attackers are using automation to discover and exploit vulnerabilities faster than humans can patch them. The mismatch between the speed of attack and the speed of defense is the defining challenge of modern application security.

This session explores how artificial intelligence is reshaping the vulnerability management lifecycle from detection through prioritization to automated remediation. We will examine how AI can distinguish critical, exploitable vulnerabilities from noise, how large language models can generate contextually aware fix suggestions, and what the real-world limitations and risks are (including hallucinated fixes and over-reliance on automation). Drawing on industry-wide trends and publicly available research, this talk provides a balanced, practitioner-focused view of where AI-powered AppSec is heading and what security teams should be doing now to prepare. No vendor pitches just an honest assessment of the opportunity, the limitations, and the path forward.

Format: Session Presentation (30 min) or Panel Discussion
Level: Intermediate

Why India's Digital Infrastructure Depends on Supply Chain Security

India is the second-largest and fastest-growing developer community on GitHub, with over 20 million developers and 30% year-over-year growth. From Aadhaar to UPI to the India Stack, the country has built world-class digital public infrastructure serving over a billion people. But this infrastructure is built on the same global open-source supply chain that is under active attack.

This session examines why software supply chain security matters disproportionately for India as both the world's largest software services exporter and a massive consumer of open-source components. We will walk through recent high-profile supply chain attacks, explain how automated tools like Dependabot protect millions of repositories globally, and discuss what Indian organizations, developers, and policymakers can do to strengthen their supply chain posture. From CERT-In's vulnerability disclosure mandates to the adoption of SBOMs and automated dependency management, this talk provides a roadmap for India's next chapter in cybersecurity.

Format: Session Presentation (30 min) best for OWASP India, NASSCOM events, or India-focused tracks

Level: Introductory

From SolarWinds to LiteLLM: A Five-Year Retrospective on Supply Chain Attacks

In five years, supply chain attacks have evolved from a niche concern to a board-level risk. This session traces the arc from SolarWinds (2020) through Log4j (2021), the NPM and PyPI dependency confusion campaigns (2022–2024), and the recent Trivy-to-LiteLLM cascading attack (2026). What patterns emerge? What defenses work? And what does the next five years look like as AI accelerates both attack and defense? A fast-paced, insight-dense session designed to give any audience technical or executive a clear picture of where we have been, where we are, and where we are heading.

Format: Lightning Talk (5–10 min) or Session Presentation (30 min)
Level: Introductory to Intermediate

Operationalizing Software Supply Chain Security at Ecosystem Scale

Modern applications are mostly code you didn't write - hundreds of transitive open-source dependencies, where a single compromised package can ripple across thousands of downstream systems. This hands-on workshop takes you from the theory of supply chain risk to the practical mechanics of defending against it at scale.

Drawing on experience running Dependabot - which monitors 30M+ repositories across 34+ package ecosystems - we walk through how dependency alerts, automated security updates, and vulnerability response work in production, and how they map to the NIST Secure Software Development Framework (SSDF). You'll triage a realistic backlog of vulnerable dependencies, prioritize remediation by severity and exploitability rather than raw CVE counts, and design an automated response workflow for your own organization - including a clear-eyed look at where AI coding agents help and where they don't.

You'll leave with a concrete, framework-aligned playbook for reducing dependency risk, not just an awareness of the problem.

Interactive workshop, co-delivered by Ankit Kumar Honey (Senior Engineering Manager, Dependabot, GitHub) and Eshaan Jain.

- Preferred duration: 90 minutes.
- Target audience: software, platform, and application-security engineers and engineering managers; intermediate level.
- Prerequisites: laptop with a code editor and Git; GitHub account helpful but not required (sample repo provided).