The Log is the Agent: Event-Sourced Reactive Graphs for Auditable, Forkable Agentic Systems

TL;DR

The paper introduces ActiveGraph, a runtime system that uses an append-only event log as the source of truth, enabling deterministic replay, cheap forking, and full lineage tracing for agent systems.

Contribution

It presents a novel architecture that inverts traditional agent frameworks by making the event log the core source, enhancing observability and reproducibility.

Findings

Deterministic replay of agent runs from logs.

Efficient forking of agent states without re-execution.

Full causal lineage from goals to model outputs.

Abstract

Most agent frameworks are built around the language model: a conversation loop comes first, then tools, then rules, and finally a logging layer bolted on for observability, with state persisted as retrievable "memory." We describe ActiveGraph, a runtime that inverts this arrangement. The append-only event log is the source of truth; the working graph is a deterministic projection of that log; and behaviors--ordinary functions, classes, LLM-backed routines, or logic attached to typed edges--react to changes in the graph and emit new events. No component instructs another; coordination happens entirely through the shared graph. This single design decision yields three properties that retrieval-and-summarization memory systems do not provide: deterministic replay of any run from its log, cheap forking that branches a run at any event without re-executing the shared prefix, and end-to-end lineage from a high-level goal down to the individual model call that produced each artifact. We present the architecture, a determinism contract that makes replay sound, and a worked diligence example whose full causal structure is reconstructable from the log alone. We discuss--without claiming to demonstrate--why this substrate is unusually well suited to self-improving agents, and how it extends the BabyAGI lineage and prior graph-memory research.