mlsanityData to 14 Oct 2026

Functional Cache Grafting: Robust and Rapid Code-Policy Synthesis for Embodied Agents

arXiv 2606.13097 ↗First posted 11 Jun 2026cs.PL cs.AIcited by 0

On the map: LLM agents and tool use → Multimodal and embodied agents

Abstract (arXiv)

Code-writing large language models (CodeLLMs) generate executable code policies for embodied agents by translating natural language goals and environmental constraints into structured control programs. However, policy generation in open-domain embodied environments suffers from two fundamental limitations: (i) delayed decoding caused by repetitive prefill computation over long prompts, and (ii) limited robustness due to fully generative decoding, which often produces API mismatches, missing safety guards, and unstable control logic. To address these limitations, we present FCGraft, a Functional Cache Grafting framework. FCGraft maintains a library of function-level validated code skeletons and their associated prompt-level Transformer key-value (KV) caches, and synthesizes new policies by retrieving relevant functions and grafting their KV caches when a new task is provided. Given retrieved function caches, FCGraft performs cache grafting via stitching, which composes cached function segments into a composite policy, and patching, which locally adapts only the necessary code regions to satisfy task-specific parameters and constraints with minimal additional decoding. By eliminating redundant prefill computation, this approach reduces generation latency, while reusing validated control structures improves robustness over prompt-level caching methods RAGCache, achieving 18.31% higher task success rate and 2.3x faster policy synthesis.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Multimodal and embodied agents, one of 16 directions in LLM agents and tool use. Builds on 49 of its 61 references that are in the map. First posted on arXiv 11 Jun 2026.

Builds on

49 of 61 references.

Closest ancestors · 4

Same direction, newest first

  1. 2024HELPER-X: A Unified Instructable Embodied Agent to Tackle Four Interactive Vision-Language Domains with Memory-Augmented Language Models29 Apr
  2. 2023Open-Ended Instructable Embodied Agents with Memory-Augmented Large Language Models23 Oct
  3. 2022LLM-Planner: Few-Shot Grounded Planning for Embodied Agents with Large Language Models8 Dec
  4. 2022Language Models as Zero-Shot Planners: Extracting Actionable Knowledge for Embodied Agents18 Jan

Other references · 45

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  2. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  3. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  4. 2023Reflexion: Language Agents with Verbal Reinforcement Learningcited by 4,276
  5. 2022Do As I Can, Not As I Say: Grounding Language in Robotic Affordancescited by 2,348
  6. 2019HuggingFace's Transformers: State-of-the-art Natural Language Processingcited by 2,218
  7. 2023Code Llama: Open Foundation Models for Codecited by 2,139
  8. 2017Gradient Episodic Memory for Continual Learningcited by 1,898
  9. 2022Code as Policies: Language Model Programs for Embodied Controlcited by 1,210
  10. 2024Qwen2.5-Coder Technical Reportcited by 1,203
  11. 2024DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligencecited by 1,181
  12. 2022Inner Monologue: Embodied Reasoning through Planning with Language Modelscited by 1,081
33 more
  1. 2025A-MEM: Agentic Memory for LLM Agentscited by 1,063
  2. 2022CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesiscited by 885
  3. 2019ALFRED: A Benchmark for Interpreting Grounded Instructions for Everyday Taskscited by 783
  4. 2023VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Modelscited by 750
  5. 2019RLBench: The Robot Learning Benchmark & Learning Environmentcited by 689
  6. 2023Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Modelscited by 574
  7. 2018Data-Efficient Hierarchical Reinforcement Learningcited by 455
  8. 2024DeepSeek-Coder-V2: Breaking the Barrier of Closed-Source Models in Code Intelligencecited by 331
  9. 2023Prompt Cache: Modular Attention Reuse for Low-Latency Inferencecited by 268
  10. 2022Efficient Training of Language Models to Fill in the Middlecited by 255
  11. 2023Large Language Models as Commonsense Knowledge for Large-Scale Task Planningcited by 231
  12. 2024CacheBlend: Fast Large Language Model Serving for RAG with Cached Knowledge Fusioncited by 223
  13. 2015Contextual Markov Decision Processescited by 220
  14. 2021TEACh: Task-driven Embodied Agents that Chatcited by 197
  15. 2025OmniDrive: A Holistic Vision-Language Dataset for Autonomous Driving with Counterfactual Reasoningcited by 143
  16. 2023Structured Chain-of-Thought Prompting for Code Generationcited by 138
  17. 2024RAGCache: Efficient Knowledge Caching for Retrieval-Augmented Generationcited by 98
  18. 2024RAP: Retrieval-Augmented Planning with Contextual Memory for Multimodal LLM Agentscited by 80
  19. 2024TurboRAG: Accelerating Retrieval-Augmented Generation with Precomputed KV Caches for Chunked Textcited by 63
  20. 2024RoboCodeX: Multimodal Code Generation for Robotic Behavior Synthesiscited by 58
  21. 2024EPIC: Efficient Position-Independent Caching for Serving Large Language Modelscited by 47
  22. 2024Don't Do RAG: When Cache-Augmented Generation is All You Need for Knowledge Taskscited by 38
  23. 2024Lifelong Robot Library Learning: Bootstrapping Composable and Generalizable Skills for Embodied Control with Language Modelscited by 27
  24. 2025Optimus-2: Multimodal Minecraft Agent with Goal-Observation-Action Conditioned Policycited by 25
  25. 2022Residual Skill Policies: Learning an Adaptable Skill-based Action Space for Reinforcement Learning for Roboticscited by 24
  26. 2025NeSyC: A Neuro-symbolic Continual Learner For Complex Embodied Tasks In Open Domainscited by 15
  27. 2024Let the Code LLM Edit Itself When You Edit the Codecited by 10
  28. 2025Mell: Memory-Efficient Large Language Model Serving via Multi-GPU KV Cache Managementcited by 8
  29. 2025MPIC: Position-Independent Multimodal Context Caching System for Efficient MLLM Servingcited by 8
  30. 2024Is Functional Correctness Enough to Evaluate Code Language Models? Exploring Diversity of Generated Codescited by 5
  31. 2024MCCoder: Streamlining Motion Control with LLM-Assisted Code Generation and Rigorous Verificationcited by 5
  32. 2024Textualized Agent-Style Reasoning for Complex Tasks by Multiple Round LLM Generationcited by 3
  33. 2025EFIM: Efficient Serving of LLMs for Infilling Tasks with Improved KV Cache Reusecited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026CaP-X: A Framework for Benchmarking and Improving Coding Agents for Robot Manipulation
  2. 2026TypeGo: An OS Runtime for Embodied Agents
  3. 2026AgenticCache: Cache-Driven Asynchronous Planning for Embodied AI Agents
  4. 2026CEDAR: Automata as Verifiable Interfaces for Language-Guided Embodied Action
  5. 2025HyCodePolicy: Hybrid Language Controllers for Multimodal Monitoring and Decision in Embodied Agents
  6. 2026Embodied Turing Machines: Stateful Code for Robot Recursive Self-Improvement
  7. 2025Runaway is Ashamed, But Helpful: On the Early-Exit Behavior of Large Language Model-based Agents in Embodied Environments
  8. 2026TurnCost: Turns, Not Observations: Cache-Aware Loop Engineering for Coding Agents