UndoBench: Decoupling Nominal Task Competence from Fault Recovery in Tool-Using AI Agents via Counterfactual Mutation Paired Trials
Tool-using AI agents are rapidly assuming mission-critical automation responsibilities across enterprise software ecosystems. However, standard agent benchmarks focus exclusively on nominal, forward task completion—conflating initial planning competency with operational resilience and fault recovery. In real-world enterprise environments where network drops, idempotency failures, and partial commits are commonplace, unguided LLM retries routinely cause devastating side effects such as duplicate payments or corrupted databases. Researchers introduce UndoBench (arXiv:2610.05622), a comprehensive evaluation suite spanning 36 enterprise workflows and 36 production fault scenarios across 8 enterprise verticals. Utilizing counterfactual paired trials alongside wire-level effect-history and state oracles, UndoBench decouples task competence from recovery capability. Experiments reveal a severe vulnerability gap: while nominal task competence achieves 83.54%, conditional recovery success rates (CRSR) collapse to 46.72%, with naive retry producing duplicated external mutations in 53.33% of trials. The code is available at GitHub (tradertanmay/undobench).
Key Takeaways
- Introduces UndoBench, the first enterprise fault recovery and undo benchmark for tool-using AI agents across 8 domains and 36 fault scenarios
- Decouples nominal task competence from fault recovery: nominal task completion hits 83.54% while conditional recovery plunges to 46.72%, with naive retries causing duplicate mutations in 53.33% of trials
- Pinpoints phase-dependent execution vulnerabilities, urging agent frameworks to move from naive retry loops to deterministic compensation transactions