{"slug": "context-as-infrastructure-knowledge-compilation-for-retrieval-in-agent-harness", "title": "Context as Infrastructure: Knowledge Compilation for Retrieval in Agent Harness", "summary": "A paper submitted to arXiv on 30 Oct 2026 reports that a search harness using only BM25/grep and MiniMax M3 answers 85% of BrowseComp-Plus and 91% on multi-hop FRAMES without compiled or vector indexes, and that a navigation tree whose leaves are claim-evidence pairs cuts query tokens by 91% when evidence suffices and adds 1.9% when it does not, a 38% average reduction. The authors, describing RAGFlow's open-source pipeline and harness, report a 2% break-even sufficiency rate against a measured rate near 43%, and a decision gate that cuts latency from 140 s to 30 s. The paper frames knowledge compilation as a supply layer rather than a better retriever, noting structural artifacts gave no detectable accuracy gain on FRAMES under a competent harness and a PageIndex-style outline degraded at scale.", "body_md": "# Computer Science > Information Retrieval\n\n  [Submitted on 30 Oct 2026]\n\n# Title:Context as Infrastructure: Knowledge Compilation for Retrieval in Agentic Systems with Harnesses\n\n[View PDF](https://arxiv.org/pdf/2610.41172)\n\n[HTML (experimental)](https://arxiv.org/html/2610.41172v1)\n\nAbstract:Retrieval-augmented generation is becoming agentic. A search harness writes queries, reads results, reformulates, and decides when to stop. This largely solves accuracy: a harness with only BM25/grep and MiniMax M3 answers 85% of BrowseComp-Plus and 91% on multi-hop FRAMES in our runs, without compiled or vector indexes. It exposes a context problem: an undisciplined loop repeats queries, rereads documents, and fills the window with unused passages. The missing layer is knowledge compilation: a configurable ingestion pipeline where LLMs enrich documents with context no chunk contains alone--navigation hierarchies with claim-evidence leaves, entity graphs, timelines, mind maps, ontologies, wikis, and memory. Compilation is not a better retriever: under a competent harness, structural artifacts gave no detectable accuracy gain on FRAMES, and a PageIndex-style outline degraded at scale. It is a supply layer. A navigation tree whose leaves are claim-evidence pairs--the mechanism we measure--cuts query tokens by 91% when evidence suffices and adds 1.9% when it does not, a 38% average reduction; this implies a 2% break-even sufficiency rate and a measured rate near 43%. A decision gate cuts latency from 140 s to 30 s, though substitution's token saving is bought with retrieval rounds and carries its own latency cost. Other artifacts--ontology, graph, wiki, timeline, mind map, memory--pay off in deployment (governance, integration, onboarding, temporal orientation, continuity) where accuracy benchmarks cannot see them. We describe RAGFlow's pipeline and harness (open source at [this https URL](https://github.com/infiniflow/ragflow)), which turns the loop into five disciplined stations and keeps the context entering the model the smallest sufficient set; together they form an agent-native database: a store that hands the agent compiled, governed context instead of a ranked list of passages.\n    \n\n### Additional Features\n\n### Current browse context:\n\ncs.IR\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/))\n# 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))\n# 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))\n# 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/context-as-infrastructure-knowledge-compilation-for-retrieval-in-agent-harness", "canonical_source": "https://arxiv.org/abs/2610.41172", "published_at": "2026-11-02 04:29:15+00:00", "updated_at": "2026-11-02 04:51:08.022087+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "ai-research", "ai-tools", "ai-infrastructure"], "entities": ["arXiv", "RAGFlow", "MiniMax M3", "BrowseComp-Plus", "FRAMES", "PageIndex", "Infiniflow"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/context-as-infrastructure-knowledge-compilation-for-retrieval-in-agent-harness", "markdown": "https://wpnews.pro/news/context-as-infrastructure-knowledge-compilation-for-retrieval-in-agent-harness.md", "text": "https://wpnews.pro/news/context-as-infrastructure-knowledge-compilation-for-retrieval-in-agent-harness.txt", "jsonld": "https://wpnews.pro/news/context-as-infrastructure-knowledge-compilation-for-retrieval-in-agent-harness.jsonld"}}