Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite
cs.AI
Submitted: 2026-10-02
Updated: 2026-10-02
Code: https://github.com/krafton-ai/kira
Sources
- Jev for Scientific Decisions: Evaluating Semantic Choices and Their Consequences
- Raven: The Harness of Harnesses for Composable Agentic Intelligence
- VisPlay: Self-Evolving Vision-Language Models from Images
- Meta-Harness: End-to-End Optimization of Model Harnesses
- Recursive Synthesis for Long-Horizon Terminal Tasks
- MM-Zero: Self-Evolving Multi-Model Vision Language Models From Zero Data
- Long-Horizon-Terminal-Bench: Testing the Limits of Agents on Long-Horizon Terminal Tasks with Dense Reward-Based Grading
- Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnesses
- SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness
- AutoHarness: improving LLM agents by automatically synthesizing a code harness
- SKILL0: In-Context Agentic Reinforcement Learning for Skill Internalization
- EvoHarness-RL: Learning Runtime Harness Coordination for Self-Evolving Agents
- Training Software Engineering Agents and Verifiers with SWE-Gym
- AI4AI at Test-Time: Strong-to-Weak Capability Transfer via Harnesses
- PATS: Policy-Aware Training Scaffolding for Agentic Reinforcement Learning
- DCAS: Decoupling CLI Agent Scaffolding to Internalize Planning across Scaffolds
- The Scaffold Effect in Coding Agents: Harness Choice as a Hidden Variable in Coding-Agent Evaluation
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