{"slug": "tool-schema-compression-enables-agentic-rag-under-constrained-context-budgets", "title": "Tool-schema compression enables agentic RAG under constrained context budgets", "summary": "Researchers have demonstrated that compressing tool schemas by 44-50% enables agentic retrieval-augmented generation (RAG) systems to function under tight context budgets, where uncompressed definitions would otherwise overflow the window and cause near-zero performance. In tests across 14 models and 6,566 API calls, compressed schemas restored exact-match accuracy by an average of 20.5 percentage points at 8K token budgets, while showing negligible impact when context was ample at 32K tokens. The findings establish tool-schema compression as essential infrastructure for deploying agentic RAG in constrained environments, with compressed schemas supporting over 800 tools compared to JSON schemas that overflow at roughly 494.", "body_md": "# Computer Science > Software Engineering\n\n[Submitted on 24 May 2026]\n\n# Title:Tool-Schema Compression Enables Agentic RAG Under Constrained Context Budgets\n\n[View PDF](/pdf/2605.26165)\n\n[HTML (experimental)](https://arxiv.org/html/2605.26165v1)\n\nAbstract:Agentic RAG systems that equip language models with dozens to hundreds of tool definitions face a critical resource conflict: tool schemas consume the same context window needed for retrieval-augmented generation. We present the first systematic study of this tool-context trade-off, evaluating 14 models spanning 1.5B-32B local models plus one frontier API model across 6,566 controlled API calls at three context budgets (8K, 16K, 32K) with 28 tool definitions. Applying TSCG conservative-profile compression (44-50% schema token savings), we observe a binary enablement effect: at 8K tokens, JSON-schema tool definitions overflow the context window entirely, yielding near-zero EM (2.6% average), while compressed schemas restore RAG functionality with +20.5 pp average exact-match lift across all eight models (+24.7 pp among the six exhibiting full enablement). At 32K -- where both formats fit -- four of five tested models show delta <= 1 pp, confirming the effect is purely budget-driven. External validation on HotpotQA (50 multi-hop questions) shows +48 pp EM under the same overflow scenario. Frontier scaling tests demonstrate that JSON schemas overflow at ~494 tools while compressed schemas remain operational beyond 800 tools. Our results establish tool-schema compression as a necessary infrastructure layer for agentic RAG in constrained-context deployments. All code, data, and checkpoints are publicly available.\n\n### Current browse context:\n\ncs.SE\n\n### References & Citations\n\nLoading...\n\n# Bibliographic and Citation Tools\n\nBibliographic Explorer\n\n*(*[What is the Explorer?](https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer))\nConnected Papers\n\n*(*[What is Connected Papers?](https://www.connectedpapers.com/about))\nLitmaps\n\n*(*[What is Litmaps?](https://www.litmaps.co/))\nscite Smart Citations\n\n*(*[What are Smart Citations?](https://www.scite.ai/))# Code, Data and Media Associated with this Article\n\nalphaXiv\n\n*(*[What is alphaXiv?](https://alphaxiv.org/))\nCatalyzeX Code Finder for Papers\n\n*(*[What is CatalyzeX?](https://www.catalyzex.com))\nDagsHub\n\n*(*[What is DagsHub?](https://dagshub.com/))\nGotit.pub\n\n*(*[What is GotitPub?](http://gotit.pub/faq))\nHugging Face\n\n*(*[What is Huggingface?](https://huggingface.co/huggingface))\nScienceCast\n\n*(*[What is ScienceCast?](https://sciencecast.org/welcome))# Demos\n\n# Recommenders and Search Tools\n\nInfluence Flower\n\n*(*[What are Influence Flowers?](https://influencemap.cmlab.dev/))\nCORE Recommender\n\n*(*[What is CORE?](https://core.ac.uk/services/recommender))# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.\n\nHave an idea for a project that will add value for arXiv's community? [ Learn more about arXivLabs](https://info.arxiv.org/labs/index.html).", "url": "https://wpnews.pro/news/tool-schema-compression-enables-agentic-rag-under-constrained-context-budgets", "canonical_source": "https://arxiv.org/abs/2605.26165", "published_at": "2026-05-27 08:05:37+00:00", "updated_at": "2026-05-27 08:49:12.483721+00:00", "lang": "en", "topics": ["large-language-models", "artificial-intelligence", "ai-agents", "natural-language-processing", "ai-research"], "entities": ["HotpotQA", "TSCG"], "alternates": {"html": "https://wpnews.pro/news/tool-schema-compression-enables-agentic-rag-under-constrained-context-budgets", "markdown": "https://wpnews.pro/news/tool-schema-compression-enables-agentic-rag-under-constrained-context-budgets.md", "text": "https://wpnews.pro/news/tool-schema-compression-enables-agentic-rag-under-constrained-context-budgets.txt", "jsonld": "https://wpnews.pro/news/tool-schema-compression-enables-agentic-rag-under-constrained-context-budgets.jsonld"}}