# Kintsugi: Learning Policies by Repairing Executable Knowledge Bases

arXiv: https://arxiv.org/abs/2605.09487 · HTML: https://mlsanity.com/paper/2605.09487/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 31 of its 36 references that are in the map.
- First posted on arXiv 10 May 2026.

## Builds on (most cited first, 30 of 31)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control](https://mlsanity.com/paper/2307.15818/) (2023)
- [OpenVLA: An Open-Source Vision-Language-Action Model](https://mlsanity.com/paper/2406.09246/) (2024)
- [Diffusion Policy: Visuomotor Policy Learning via Action Diffusion](https://mlsanity.com/paper/2303.04137/) (2023)
- [π0: A Vision-Language-Action Flow Model for General Robot Control](https://mlsanity.com/paper/2410.24164/) (2024)
- [Do As I Can, Not As I Say: Grounding Language in Robotic Affordances](https://mlsanity.com/paper/2204.01691/) (2022)
- [Voyager: An Open-Ended Embodied Agent with Large Language Models](https://mlsanity.com/paper/2305.16291/) (2023)
- [Octo: An Open-Source Generalist Robot Policy](https://mlsanity.com/paper/2405.12213/) (2024)
- [Mastering Diverse Domains through World Models](https://mlsanity.com/paper/2301.04104/) (2023)
- [WebShop: Towards Scalable Real-World Web Interaction with Grounded Language Agents](https://mlsanity.com/paper/2207.01206/) (2022)
- [Code as Policies: Language Model Programs for Embodied Control](https://mlsanity.com/paper/2209.07753/) (2022)
- [ALFWorld: Aligning Text and Embodied Environments for Interactive Learning](https://mlsanity.com/paper/2010.03768/) (2020)
- [Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learning](https://mlsanity.com/paper/1910.10897/) (2019)
- [ExpeL: LLM Agents Are Experiential Learners](https://mlsanity.com/paper/2308.10144/) (2023)
- [TD-MPC2: Scalable, Robust World Models for Continuous Control](https://mlsanity.com/paper/2310.16828/) (2023)
- [robosuite: A Modular Simulation Framework and Benchmark for Robot Learning](https://mlsanity.com/paper/2009.12293/) (2020)
- [Cognitive Architectures for Language Agents](https://mlsanity.com/paper/2309.02427/) (2023)
- [Agent Workflow Memory](https://mlsanity.com/paper/2409.07429/) (2024)
- [Integrated Task and Motion Planning](https://mlsanity.com/paper/2010.01083/) (2020)
- [ADaPT: As-Needed Decomposition and Planning with Language Models](https://mlsanity.com/paper/2311.05772/) (2023)
- [PDDLStream: Integrating Symbolic Planners and Blackbox Samplers via Optimistic Adaptive Planning](https://mlsanity.com/paper/1802.08705/) (2018)
- [π0.7: a Steerable Generalist Robotic Foundation Model with Emergent Capabilities](https://mlsanity.com/paper/2604.15483/) (2026)
- [Learning Symbolic Operators for Task and Motion Planning](https://mlsanity.com/paper/2103.00589/) (2021)
- [Predicate Invention for Bilevel Planning](https://mlsanity.com/paper/2203.09634/) (2022)
- [Learning Neuro-Symbolic Relational Transition Models for Bilevel Planning](https://mlsanity.com/paper/2105.14074/) (2021)
- [VisualPredicator: Learning Abstract World Models with Neuro-Symbolic Predicates for Robot Planning](https://mlsanity.com/paper/2410.23156/) (2024)
- [InterPreT: Interactive Predicate Learning from Language Feedback for Generalizable Task Planning](https://mlsanity.com/paper/2405.19758/) (2024)
- [StateAct: Enhancing LLM Base Agents via Self-prompting and State-tracking](https://mlsanity.com/paper/2410.02810/) (2024)
- [Embodied Lifelong Learning for Task and Motion Planning](https://mlsanity.com/paper/2307.06870/) (2023)
- [ExoPredicator: Learning Abstract Models of Dynamic Worlds for Robot Planning](https://mlsanity.com/paper/2509.26255/) (2025)

## Built on by (most cited first, 3 of 3)

- [Learning Explicit Behavioral Models with Adaptive Questions and World-Model Probes](https://mlsanity.com/paper/2606.07127/) (2026)
- [From Agent Failures to Text Policies: What Works and What Breaks](https://mlsanity.com/paper/2607.20668/) (2026)
- [ALDER: Discovering the Laws of a World by Acting in It](https://mlsanity.com/paper/2609.33728/) (2026)

## Abstract (arXiv)

Modern embodied agents achieve impressive performance, but their task knowledge is often stored in neural weights, latent state, or prompt-bound memory, making individual policy knowledge difficult to inspect, validate, recombine, and reuse. We introduce Kintsugi, a white-box policy-learning framework that treats embodied policy improvement as verifier-gated construction of a typed executable Knowledge Base (KB). Kintsugi represents task-level policy knowledge as composable typed entries -- predicates, operators, policy schemas, monitors, recovery rules, experience records, and goals -- and improves this artifact through localized typed edits induced from rollout evidence, rather than relying on test-time language-model reasoning. Between rollouts, a tool-constrained agentic editing loop diagnoses trajectory failures, localizes them to editable KB layers, and proposes candidate edits. A deterministic verification gate admits an edit only when the candidate type-checks, the resulting KB executes, and focused validation success or trajectory-health metrics improve without violating protected-regression checks. At inference, the accepted KB is executed by a deterministic symbolic executor with zero LLM calls. Across long-horizon text-agent benchmarks and representative object-centric manipulation settings, Kintsugi achieves strong endpoint performance while preserving inspectability, local editability, and verifier-gated deployment. These results suggest that embodied policy improvement can be organized around executable task knowledge.