LLM Agents & Planning: Literature Digest

Large language model agents planning: A literature digest

Recent work on large language model (LLM) agents planning has moved beyond single-step prompting toward richer, system‑embedded planning architectures. LLMs are increasingly treated as planners within broader agent systems, supporting task decomposition, tool use, multi‑agent coordination, and grounded, interactive planning.

Grounded and interactive planning

Several papers emphasize grounded, interactive planning where LLMs map natural language instructions to environment‑constrained actions. LLM‑Planner: Few‑shot grounded planning for embodied agents with large language models positions LLMs as planners for embodied agents that must follow natural language while respecting environmental constraints. Describe, explain, plan and select: Interactive planning with large language models enables open‑world multi‑task agents introduces DEPS, an interactive framework in which an LLM acts as a zero‑shot planner that iteratively describes, explains, plans, and selects actions. TPTU: Task planning and tool usage of large language model‑based AI agents focuses on how LLMs can plan the order of tool usage, evaluating their ability to sequence tools effectively for complex tasks.

Multi‑agent and systematic perspectives

Other works situate LLM planning in multi‑agent and broader planning contexts. Large language model based multi‑agents: A survey of progress and challenges surveys how LLMs support planning in multi‑agent systems, highlighting progress and open issues. Smart‑LLM: Smart multi‑agent robot task planning using large language models uses LLMs to decompose and coordinate robot tasks in multi‑agent settings. On the prospects of incorporating large language models (LLMs) in automated planning and scheduling (APS) discusses how LLMs can complement classical APS methods, noting gaps in multi‑agent communication and belief‑state management. Large language models for planning: A comprehensive and systematic survey provides a high‑level taxonomy of LLM‑based planning, organizing work into task decomposition, plan selection, external modules, and reflection/memory.

Open problems

• Standardizing inter‑agent communication and belief‑state management in multi‑agent LLM planning.
• Improving robustness of LLM‑generated plans under partial observability and noisy environments.
• Bridging LLM‑based planning with formal automated planning and scheduling guarantees.
• Evaluating LLM planning abilities systematically across diverse, commonsense‑rich domains.
• Scaling interactive, grounded planning to long‑horizon, multi‑task embodied agents.

Key papers

1. Llm-planner: Few-shot grounded planning for embodied agents with large language models — CH Song,J Wu,C Washington…
2. Large language model based multi-agents: A survey of progress and challenges — T Guo,X Chen,Y Wang,R Chang,S Pei…
3. On the planning abilities of large language models-a critical investigation — K Valmeekam,M Marquez…
4. Describe, explain, plan and select: Interactive planning with large language models enables open-world multi-task agents — Z Wang,S Cai,G Chen,A Liu,X Ma,Y Liang
5. Tptu: Task planning and tool usage of large language model-based ai agents — J Ruan,Y Chen,B Zhang,Z Xu,T Bao…
6. Smart-llm: Smart multi-agent robot task planning using large language models — SS Kannan,VLN Venkatesh…
7. On the prospects of incorporating large language models (llms) in automated planning and scheduling (aps) — V Pallagani,BC Muppasani,K Roy,F Fabiano…
8. Large language models for planning: A comprehensive and systematic survey — P Cao,T Men,W Liu,J Zhang,X Li,X Lin,D Sui…
9. Embodied task planning with large language models — Z Wu,Z Wang,X Xu,J Lu,H Yan
10. AgentGen: Enhancing Planning Abilities for Large Language Model based Agent via Environment and Task Generation — M Hu,P Zhao,C Xu,Q Sun,JG Lou,Q Lin…

Papers via the AISA Scholar API; synthesis by the AISA LLM layer. 2026-06-23.