AI
WideSWE: Can Coding Agents Coordinate Changes Across Repositories?
Publication: Hugging Face Daily Papers
Author: Not available in collected metadata
Source published: 26/09/2026, 20:00:00 UTC
Briefing created: 29/09/2026, 12:28:48 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 29/09/2026, 06:18:24 UTC.
Briefings
This study introduces WideSWE to evaluate coding agents on cross-repository tasks, addressing the limitation where evaluation remains confined to single codebases. The research analyzes 120 real-world tasks across 103 software ecosystems, comparing independent versus joint execution strategies. Key findings indicate that joint execution significantly outperforms independent execution by leveraging information from related repositories to guide implementation and verification, with the Codex CLI + GPT-5.6-sol configuration achieving the highest success rate of 42.50%.
Key points
- The study introduces WideSWE to evaluate coding agents on cross-repository tasks, addressing the limitation where evaluation remains confined to single codebases.
- Joint execution significantly outperforms independent execution by leveraging information from related repositories to guide implementation and verification, with the Codex CLI + GPT-5.6-sol configuration achieving the highest success rate of 42.50%.
Context and limitations
The evaluation covers 103 software ecosystems with 120 tasks, but the scope is limited to software ecosystems rather than all software domains. Results are based on real-world tasks derived from mining and reviewing changes, and the study compares independent versus joint execution strategies under identical prompts.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 29/09/2026, 12:28:48 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeiaffphrnjcikngtydtcrbaxfzsuc5lsmzwprdmomuaj7b5hethg6a