mlsanityData to 14 Oct 2026

Dynamic System Instructions and Tool Exposure for Efficient Agentic LLMs

arXiv 2602.17046 ↗First posted 1 Dec 2025cs.AIcited by 3

On the map: LLM agents and tool use → Tool learning and function calling

Abstract (arXiv)

Large Language Model (LLM) agents often run for many steps while re-ingesting long system instructions and large tool catalogs each turn. This increases cost, agent derailment probability, latency, and tool-selection errors. We propose Instruction-Tool Retrieval (ITR), a RAG variant that retrieves, per step, only the minimal system-prompt fragments and the smallest necessary subset of tools. ITR composes a dynamic runtime system prompt and exposes a narrowed toolset with confidence-gated fallbacks. Using a controlled benchmark with internally consistent numbers, ITR reduces per-step context tokens by 95%, improves correct tool routing by 32% relative, and cuts end-to-end episode cost by 70% versus a monolithic baseline. These savings enable agents to run 2-20x more loops within context limits. Savings compound with the number of agent steps, making ITR particularly valuable for long-running autonomous agents. We detail the method, evaluation protocol, ablations, and operational guidance for practical deployment.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Placed in Tool learning and function calling, one of 16 directions in LLM agents and tool use. Builds on 26 of its 36 references that are in the map. First posted on arXiv 1 Dec 2025.

Builds on

26 of 36 references.

Closest ancestors · 4

Same direction, newest first

  1. 2025ScaleMCP: Dynamic and Auto-Synchronizing Model Context Protocol Tools for LLM Agents9 May
  2. 2023ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs31 Jul
  3. 2023Gorilla: Large Language Model Connected with Massive APIs24 May
  4. 2023Toolformer: Language Models Can Teach Themselves to Use Tools9 Feb

Other references · 22

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  3. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  4. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,455
  5. 2023SWE-bench: Can Language Models Resolve Real-World GitHub Issues?cited by 3,373
  6. 2023A Survey on Large Language Model based Autonomous Agentscited by 2,242
  7. 2023Retrieval-Augmented Generation for Large Language Models: A Surveycited by 2,212
  8. 2023MetaGPT: Meta Programming for A Multi-Agent Collaborative Frameworkcited by 1,808
  9. 2021WebGPT: Browser-assisted question-answering with human feedbackcited by 1,745
  10. 2024SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineeringcited by 1,737
  11. 2020Leveraging Passage Retrieval with Generative Models for Open Domain Question Answeringcited by 1,210
  12. 2023GAIA: a benchmark for General AI Assistantscited by 1,128
10 more
  1. 2023Learning to Compress Prompts with Gist Tokenscited by 390
  2. 2023A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaborationcited by 290
  3. 2023LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Modelscited by 287
  4. 2023Unleashing the Emergent Cognitive Synergy in Large Language Models: A Task-Solving Agent through Multi-Persona Self-Collaborationcited by 251
  5. 2024Token-Budget-Aware LLM Reasoningcited by 247
  6. 2023Agents: An Open-source Framework for Autonomous Language Agentscited by 116
  7. 2021The Perils of Using Mechanical Turk to Evaluate Open-Ended Text Generationcited by 97
  8. 2022Prompt Compression and Contrastive Conditioning for Controllability and Toxicity Reduction in Language Modelscited by 97
  9. 2024Towards Optimizing the Costs of LLM Usagecited by 33
  10. 2024TRIM: Token Reduction and Inference Modeling for Cost-Effective Language Generationcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026The Workload-Router-Pool Architecture for LLM Inference Optimization: A Vision Paper from the vLLM Semantic Router Projectcited by 1
  2. 2026Task-Aware Harness Provisioning for LLM Agents in Mission-Critical Infrastructure Operationscited by 1
  3. 2026Toollery: Scaling LLM Agents to Thousands of Skills and Toolscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025MemTool: Optimizing Short-Term Memory Management for Dynamic Tool Calling in LLM Agent Multi-Turn Conversations
  2. 2026DOCSCHISEL: Adaptive Tool Documentation Optimization Framework for LLM Agents
  3. 2025PLAY2PROMPT: Zero-shot Tool Instruction Optimization for LLM Agents via Tool Play
  4. 2025AGENTIF: Benchmarking Instruction Following of Large Language Models in Agentic Scenarios
  5. 2026ASA: Backbone-Training-Free Representation Engineering for Tool-Calling Agents
  6. 2026The Evolution of Tool Use in LLM Agents: From Single-Tool Call to Multi-Tool Orchestration
  7. 2026From Atomic Actions to Standard Operating Procedures: Iterative Tool Optimization for Self-Evolving LLM Agents
  8. 2026ToolAtlas: Learning Once, Reusing Everywhere with Tool-Side Memory