{ "schema_version": 1, "builder_plan": { "scenario": "A product assistant retrieves plausible documents but gives unsupported answers.", "change": "Change one retrieval or ranking step while keeping the document snapshot fixed.", "baseline": "Your current retrieval pipeline with identical questions and answer model.", "steps": [ "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.", "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.", "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.", "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test." ], "measure": "Supported answers, retrieval misses, correct abstentions, index cost, and response latency.", "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation." }, "builder_review": null, "arxiv_id": "2610.23480", "title": "MuninAI: A Self-Hosted Agentic Framework for Independent Research Groups", "summary": "It replaces both hosted literature assistants that require sending unpublished work to third parties and naive fixed top-k retrieve-then-answer pipelines, which the paper reports score below the unaided model. Costs include serving an open-weight model locally, building and maintaining the corpus and citation-graph index, and operating the agent harness with its tool and reading components, which adds engineering complexity beyond plain RAG. It can fail on corpora or question types different from the evaluation set, the abstention behavior means a substantial fraction of out-of-corpus questions go unanswered, and gains depend on the harness rather than the model, so a poorly configured deployment may not reproduce the reported accuracy.", "rationale": "The framework is open-source, runs on a single 24 GB GPU with a 20B open-weight model, keeps data local, and ships a one-command evaluation suite the team can rerun on its own corpus before committing.", "technique": "Self-hosted agentic RAG harness for literature-grounded assistants", "family": "recuperacao_e_rag", "recommendation": "adotar", "published": "2026-10-16", "as_of": "2026-10-19", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.23480", "independent_implementations": 1, "total_implementations": 2, "stars": 132, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "It replaces both hosted literature assistants that require sending unpublished work to third parties and naive fixed top-k retrieve-then-answer pipelines, which the paper reports score below the unaided model. Costs include serving an open-weight model locally, building and maintaining the corpus and citation-graph index, and operating the agent harness with its tool and reading components, which adds engineering complexity beyond plain RAG. It can fail on corpora or question types different from the evaluation set, the abstention behavior means a substantial fraction of out-of-corpus questions go unanswered, and gains depend on the harness rather than the model, so a poorly configured deployment may not reproduce the reported accuracy.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.23480", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "On 199 paired literature questions, the agent harness raises accuracy from 0.387 to 0.874 (+0.487, 95% CI [0.407, 0.568]) relative to the unaided base model, consistent across three backbones including a 20B model on a single consumer GPU.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.23480", "source_page": null, "source_excerpt": "" } ], "exposure_map": [ { "dimension": "quality", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compute", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "latency", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "operations", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "compatibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "security", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "data_and_training", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" }, { "dimension": "reproducibility", "basis": "not_evaluated", "finding": "", "interpretation": "", "limitation": "", "next_check": "", "source_url": "", "source_page": null, "source_excerpt": "" } ], "risks": [], "minimum_test": [], "open_questions": [], "independent_tests": [], "lab_reproductions": [], "report_available": false, "validation_tier": "unknown", "evidence_tier": "unknown", "infrastructure_basis": "unknown", "equations": [], "equations_status": "not_fetched", "core_kind": "", "equations_fetched_at": "", "equations_source_url": "", "equations_source_sha256": "", "exposures_reviewed_at": "", "exposures_source_sha256": "" }