{
  "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": "2610.02952",
  "title": "GTDD: Generative Test-Driven Development for AI Coding Agents with Adversarial Testing",
  "summary": "Replaces a fixed, pre-generated test suite for AI coding agents with a separate testing agent that adaptively generates new inputs after each candidate implementation, backed by a trusted evaluator, counterexample reduction, and regression archiving. Costs additional API calls per development round, a separate tester agent, an evaluator implementation, and more complex orchestration than single-pass test generation. Can fail when the behavioral contract is underspecified, when the testing agent shares blind spots with the coding model, or when the single-task empirical evidence does not transfer to other workloads.",
  "rationale": "The method requires only API-level access and modest orchestration, so it is feasible on constrained infrastructure, but its benefit is demonstrated on one stateful task and must be validated on the reader's own coding-agent workloads before production adoption.",
  "technique": "Generative test-driven development with adversarial testing agent",
  "family": "agentes_de_codigo",
  "recommendation": "testar",
  "published": "2026-10-02",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.02952",
  "independent_implementations": 0,
  "total_implementations": 0,
  "stars": 0,
  "citations": null,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "Replaces a fixed, pre-generated test suite for AI coding agents with a separate testing agent that adaptively generates new inputs after each candidate implementation, backed by a trusted evaluator, counterexample reduction, and regression archiving. Costs additional API calls per development round, a separate tester agent, an evaluator implementation, and more complex orchestration than single-pass test generation. Can fail when the behavioral contract is underspecified, when the testing agent shares blind spots with the coding model, or when the single-task empirical evidence does not transfer to other workloads.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.02952",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "In a paired key-value-store experiment, both policies that regenerated tests during development achieved lower mean failure rates than a policy with tests generated once; no multiplicative factor is reported.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.02952",
      "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": ""
}