GitHub Blog Author Shares 10 Technical Talks for GitHub Universe 2026

The GitHub Blog features a curated list of 10 technical talks for GitHub Universe 2026, selected by an author to address practical challenges in AI agent development and infrastructure. The agenda focuses on verifying agent-written code, managing context, and securing dependencies.

Agent Memory and Evaluation

Sessions include Cooper Nederhood and Alejandro Carderera from GitHub discussing how accumulated context can degrade performance, using benchmarks from real pull requests. Christopher Harrison from GitHub will break down how to distribute context consistently across teams. Walker Chabbott from GitHub and Julia Kasper from Microsoft address the gap between benchmark scores and developer satisfaction in Copilot. Jeff An from Momentic explores how agents verify AI-generated code, while Achin Gupta from Intuit and Divya Mahajan from Amazon present an architecture where deterministic code handles signal collection and scoring, with language models narrating the evidence.

Security and Infrastructure

Security talks feature Karen Li and Leo Balter from GitHub tracing npm dependencies and provenance. Steve Glass and Greg Ose from GitHub map supply chain attack techniques to GitHub Actions controls. Nick Taylor from Pomerium demonstrates an identity-aware proxy for enforcing fine-grained authorization on hosted MCP servers.

Tooling, Connectivity, and Skills

Alexander Lichter from VoidZero presents Vite+, an open-source CLI for the front-end toolchain. Alex Junior Antwi from Braveon AI shares lessons from building CarbonSight for low-connectivity communities in Ghana. Additionally, the author joins Shishir Tewari from Procore for a session on building reusable agent skills. The event takes place October 28–29, with in-person and virtual passes available.