SWE-Critix: Post-Training Generalist Verifiers for Assessing Coding-Agent Trajectories
cs.AI
Submitted: 2026-10-09
Updated: 2026-10-12
Code: https://github.com/deepseek-ai/deepseek-harness
Terminology
Sources
- SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?
- LoLBench: Evaluating Coding Agents with Long-Horizon Proposals on Large Software Systems
- Agentless: Demystifying LLM-based Software Engineering Agents
- LLM-as-a-Verifier: A General-Purpose Verification Framework
- Trae Agent: An LLM-based Agent for Software Engineering with Test-time Scaling
- Agentic Rubrics as Contextual Verifiers for SWE Agents
- When Agents go Astray: Course-Correcting SWE Agents with PRMs
- Wink: Recovering from Misbehaviors in Coding Agents
- TRAJDEBUG: Tracing Error Lifecycle to Identify Critical Failures in Long-Horizon Agent Trajectories
- SWE-Prime: Fewer Trajectories, Better Performance
- ProcCtrlBench: Evaluating Process-Level Defects and Control Preservation in LLM Coding Agents
- Independent Patch Verification for Coding Agents with a Bidirectional Reconstruct-and-Verify Framework
- Online Monitoring and Corrective Steering of Programming Agents
- AgentLens: Revealing The Lucky Pass Problem in SWE-Agent Evaluation
- Training Software Engineering Agents and Verifiers with SWE-Gym
- R2E-Gym: Procedural Environments and Hybrid Verifiers for Scaling Open-Weights SWE Agents
- SWE-RM: Execution-free Feedback For Software Engineering Agents
- Dockerless: Environment-Free Program Verifier for Coding Agents
- SWE-TRACE: Optimizing Long-Horizon SWE Agents Through Rubric Process Reward Models and Heuristic Test-Time Scaling
- Guided Search Strategies in Non-Serializable Environments with Applications to Software Engineering Agents
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems