FlyCatcher: Neural Inference of Runtime Checkers from Tests

Beatriz Souza, Chang Lou, Suman Nath, Michael Pradel · 2026

Read FlyCatcher: Neural Inference of Runtime Checkers from Tests by Beatriz Souza, Chang Lou, Suman Nath, Michael Pradel (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-22028.txt.

Description

Souza and three colleagues present FlyCatcher, an automated way to derive runtime checkers from tests. FlyCatcher generalizes checks encoded in a test case to any run. It combines LLM-based synthesis, static analysis, plus dynamic validation, inferring checkers that monitor method calls and assert properties that should hold when called. Evaluation across 400 tests spanning four widely used, complex software systems infers 334 checkers, of which 300 are correct via cross-validation. Against a leading approach, FlyCatcher infers about 3x more correct checkers, enabling detection of 5.2x more errors.

Catalog record

CREATORBeatriz Souza, Chang Lou, Suman Nath, Michael Pradel
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE66668
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.22028

Browse the full library · About the Tower of Babel