Mon 15 Jun 2026 14:20 - 14:40 at Meadows B - PAgE Session 2

Language model (LM) agent harnesses are stateful programs around model calls: they maintain context, dispatch tools, coordinate subagents, and enforce budgets. In many implementations, the state that drives these decisions is split across control flow, in-memory caches, tool side effects, and after-the-fact logs. We present an event-driven programming model in which a durable, append-only event log is the source of truth for the state of the harness. Concerns within the harness are written as facets: independent subsystems that derive local state by pure folds over the log, authorize new events through an admission rule, and propose follow-on events when their local state changes. The log is therefore both the source of truth for facet states and the coordination medium between facets; an external trigger reactively drives a sequence of admission and proposal steps. We present this compositional design, show how it decouples harness concerns, and supports extensibility through facets. We provide a concise Python implementation of our system.

Mon 15 Jun

Displayed time zone: Mountain Time (US & Canada) change

13:40 - 15:20
PAgE Session 2PAgE at Meadows B
13:40
20m
Talk
Combining Agentic AI and Lightweight Formal Methods To Find Bugs in a Production Hypervisor
PAgE
Hiroyuki Katsura University of Cambridge, Kayvan Memarian University of Cambridge, Peter Sewell University of Cambridge
14:00
20m
Talk
From Workarounds to Root Causes: Experience Using Agentic Workflows to Debug Complex Browser GPU Compiler Stacks
PAgE
Abhijit Ramesh UC Santa Cruz, Reese Levine University of California at Santa Cruz, Tyler Sorensen University of California at Santa Cruz
14:20
20m
Talk
Event-based Design Abstractions for Agent HarnessesRecorded
PAgE
McCoy Becker , Matin Ghavami Massachusetts Institute of Technology, Fabian Zaiser Massachusetts Institute of Technology, Timothy O'Donnell , Martin C. Rinard Massachusetts Institute of Technology, Joshua B. Tenenbaum Massachusetts Institute of Technology, Vikash Mansinghka Massachusetts Institute of Technology
14:40
20m
Talk
Lumos: Let there be Language Model System CertificationRemote
PAgE
Isha Chaudhary , Vedaant Jain , Prineet Parhar , Kavya Sachdeva , Avaljot Singh University of Illinois Urbana-Champaign, Sayan Ranu , Gagandeep Singh University of Illinois Urbana-Champaign
Pre-print
15:00
20m
Talk
Lazy Validation and Self-Healing for Agentic ProgramsRemote
PAgE
Theodoros Tsampouris Aristotle University of Thessaloniki, Eleftherios Ioannidis Microsoft Research, Andreas Symeonidis Aristotle University of Thessaloniki