{ "schema_version": 1, "builder_plan": { "scenario": "A workflow needs specialist review, but extra agents may add more cost than useful work.", "change": "Test the proposed division of work against a single-agent implementation.", "baseline": "One agent with the same tools and an equal total token and time budget.", "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": "Accepted outputs, coordination failures, wall-clock time, and total cost across all agents.", "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": "2610.19999", "title": "A Strong Harness Narrows the Model Gap: A Comparative Evaluation for Legislative Document-to-XML Conversion", "summary": "It replaces reliance on a stronger foundation model with deterministic tooling, task decomposition, and persisted per-agent state wrapped around a cheaper model. The cost is engineering complexity: building and maintaining the tool layer, orchestration logic, and schema validation rather than paying for a larger model. It can fail where the residual error lives outside the tools' reach, and the pipeline may not transfer to tasks lacking strict schemas or external authorities for deterministic lookup.", "rationale": "For a team on a 24 GB GPU or paid APIs, investing in harness engineering around a mid-tier model is cheaper and more effective than scaling the model, provided the task admits decomposition and deterministic verification.", "technique": "Decomposed multi-agent harness with deterministic tools", "family": "orquestracao_multiagente", "recommendation": "adotar", "published": "2026-10-14", "as_of": "2026-10-18", "editorial_status": "indexed", "source_url": "https://arxiv.org/abs/2610.19999", "independent_implementations": 0, "total_implementations": 0, "stars": 0, "citations": null, "claims": [ { "claim_id": "claim-01", "statement": "It replaces reliance on a stronger foundation model with deterministic tooling, task decomposition, and persisted per-agent state wrapped around a cheaper model. The cost is engineering complexity: building and maintaining the tool layer, orchestration logic, and schema validation rather than paying for a larger model. It can fail where the residual error lives outside the tools' reach, and the pipeline may not transfer to tasks lacking strict schemas or external authorities for deterministic lookup.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.19999", "source_page": null, "source_excerpt": "" }, { "claim_id": "claim-02", "statement": "Weakest model in the full pipeline scores 0.948 versus 0.806 for the strongest model zero-shot (paired bootstrap p < 0.001); model spread across four dimensions narrows from 0.68-0.81 at zero-shot to ~0.95 with the full harness.", "basis": "inferred", "result": "", "baseline": "", "conditions": "", "source_url": "https://arxiv.org/abs/2610.19999", "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": "" }