{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "Your coding assistant makes changes that still need substantial repair before merging.",
    "change": "Try the technique on held-out issues from a repository with reliable tests.",
    "baseline": "The same coding agent, repository snapshot, tools, and token 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": "Correct patches, regression failures, human repair time, and total cost per accepted change.",
    "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": "2609.39982",
  "title": "Mid-Harness: Scaling Actions Between Model and Harness for Terminal Agents",
  "summary": "The method inserts a sample-and-verify step between the model and the execution harness, replacing the default practice of executing the first sampled action directly; it leaves the generator and harness unchanged. It costs additional inference per step (multiple sampled candidate actions plus verifier calls, which for the strongest reported result use a paid frontier-model API), adding per-action latency and token spend. It can fail when verification is weak: with a low-capability verifier, extra action sampling yields little benefit, so the gain depends on having a verifier stronger than the generator, and the reported results center on terminal benchmarks rather than general coding workloads.",
  "rationale": "The technique is implementable on constrained infrastructure as an inference-time wrapper with no training, but its benefit depends on verifier quality and workload, so it warrants a small-scale validation on the reader's own terminal-agent tasks before production adoption.",
  "technique": "Mid-harness action sampling and verification",
  "family": "agentes_de_codigo",
  "recommendation": "testar",
  "published": "2026-09-30",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2609.39982",
  "independent_implementations": 3,
  "total_implementations": 4,
  "stars": 131,
  "citations": null,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "The method inserts a sample-and-verify step between the model and the execution harness, replacing the default practice of executing the first sampled action directly; it leaves the generator and harness unchanged. It costs additional inference per step (multiple sampled candidate actions plus verifier calls, which for the strongest reported result use a paid frontier-model API), adding per-action latency and token spend. It can fail when verification is weak: with a low-capability verifier, extra action sampling yields little benefit, so the gain depends on having a verifier stronger than the generator, and the reported results center on terminal benchmarks rather than general coding workloads.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.39982",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "On TerminalBench-Lite, a GPT-5.6 Sol verifier raises Pass@1 from 50.00% to 68.03% with 8 sampled actions; combining action and trajectory scaling achieves higher success at lower estimated token cost than trajectory scaling alone.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2609.39982",
      "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": ""
}