Software World Models: From Consequence Prediction to Decision Value

2026-10-04

AI Summary

This study addresses the challenge that code changes can silently break downstream dependencies, while exhaustive integration testing remains impractical. To overcome this limitation, this work proposes a software world model that predicts the blast radius and decision value of changes through three stages: exploration, learning, and action. Transcending traditional static analysis, the method integrates system state recovery with candidate change execution to obtain dynamic feedback, which is subsequently used to fine-tune large language models for accurately predicting coupling failures overlooked by dependency graphs. Experimental results demonstrate that the proposed approach improves the blast radius F1 score to 0.571, halves the ranking regret, and yields significant gains in cross-project transferability.

Abstract

A coding agent may safely modify one repository while silently breaking downstream services, libraries, or datastores that depend on it. Exhaustively running integration tests after every agent action is impractical, so the agent must predict these failures before executing them. Existing software world models predict the agent's own observations, while static change-impact analysis only identifies where a change may propagate. We instead introduce the Software World Model (SWM), which models the broader system affected by a code change and predicts its blast set: the components that the change will break. SWM follows three stages: explore, learn, and act. Explore executes candidate changes from restored system states, prioritizing regions where observed failures contradict the dependency graph. Learn fine-tunes a language model on these execution outcomes to predict downstream breakage. Act converts sampled predictions into per-consumer break probabilities for change ranking, proactive migration, and deciding when another execution is worth its cost. On held-out synthetic systems, SWM improves blast-set F1 from 0.431 for static reachability to $0.571\pm0.037$, more than halves ranking regret, and improves migration return at all nine evaluation checkpoints. The two methods are complementary: reachability is stronger on dependencies represented in the graph, while SWM recovers failures caused by couplings the graph misses. Experiments on held-out real libraries further show that predicting structured failure outcomes, rather than only scalar risk, is important for downstream decision quality.

Problem

• Software World Model
• consequence prediction
• change-impact analysis
• blast set prediction
• downstream breakage

Innovation

• Software World Model
• Blast Set Prediction
• Coding Agent
• Change-Impact Analysis
• Language Model Fine-tuning