Track new results across pure and applied mathematics.
Search papers, labs, and topics across Lattice.
This paper introduces a Go-Explore-style search mechanism that leverages full sandbox snapshots to enable coding agents to recover from failures by restoring and branching from intermediate environment states. The authors demonstrate that this approach significantly improves solve rates and efficiency, particularly for complex coding tasks involving external dependencies like databases and services, where state cannot be easily reconstructed from code diffs alone. The key finding is that full-snapshot search is most beneficial when long-horizon coding efforts involve multiple interconnected services whose runtime state is critical for progress.
Coding agents can achieve higher solve rates by restoring and branching from full sandbox snapshots, especially when progress relies on external service states.
Coding agents often make useful progress before they fail, but retry-from-scratch discards the machine state created along the way. This makes failure recovery an interesting search problem: if intermediate environment states preserve meaningful progress, restoring and branching from them could outperform repeatedly restarting from the initial state. We implement Go-Explore-style search over full Daytona snapshots and find that higher and more efficient solve rates depend mainly on progress that sits outside the repository, such as databases, services, and other runtime state. Our results suggest that full-snapshot search is most effective when long-horizon coding work spans multiple services whose state cannot be reconstructed from a single diff.