{
  "schema_version": "1.0",
  "id": "s13:https://arxiv.org/abs/2607.22471v1",
  "slug": "2607-22471v1-0kivszm",
  "url": "https://feed7.dev/p/2607-22471v1-0kivszm",
  "title": "MineValiCoder: Reliable Code Generation with Test Case Quality Mining and Bipartite Graph-Based Mutual Validation",
  "why_included": "MineValiCoder filters generated tests, refines several code candidates, then scores code and tests together. The pattern targets a core agent risk: unreliable tests steering implementation.",
  "summary": "MineValiCoder links three modules: **TCQM, Parallel TDD Refinement, and BiCoTeV**. They filter faulty generated tests, refine multiple code candidates, and score code-test interactions through a bipartite graph.",
  "practical_implication": "For coding-agent harnesses, treat generated tests as uncertain evidence rather than ground truth. Validate tests before feeding failures back into implementation, preserve multiple candidates, and select using agreement across code and test behavior.",
  "agent_context": "MineValiCoder links three modules: **TCQM, Parallel TDD Refinement, and BiCoTeV**. They filter faulty generated tests, refine multiple code candidates, and score code-test interactions through a bipartite graph.\n\nFor coding-agent harnesses, treat generated tests as uncertain evidence rather than ground truth. Validate tests before feeding failures back into implementation, preserve multiple candidates, and select using agreement across code and test behavior.\n\nAcross four LLMs, reported Pass@1 reached **96.34% on HumanEval** and **51.33% on LiveCodeBench**, with intermediate results on MBPP and APPS. These benchmark scores do not establish how the added loop performs on long-horizon repository changes.",
  "source": {
    "name": "arXiv",
    "url": "https://arxiv.org/abs/2607.22471v1",
    "published_at": "2026-07-24T16:39:06.000Z"
  },
  "source_class": "blog_post",
  "content_type": "Paper",
  "layer": "agent",
  "domains": [
    "coding"
  ],
  "topics": [
    "harness-engineering",
    "agent-reliability",
    "coding-agents"
  ],
  "verification": {
    "status": "needs_review",
    "label": "Needs Review",
    "method": "unverified",
    "verified_at": null
  },
  "uncertainty": [
    "Across four LLMs, reported Pass@1 reached **96.34% on HumanEval** and **51.33% on LiveCodeBench**, with intermediate results on MBPP and APPS. These benchmark scores do not establish how the added loop performs on long-horizon repository changes."
  ],
  "lifecycle": "Current",
  "published_at": "2026-07-24T16:39:06.000Z",
  "modified_at": "2026-07-24T16:39:06.000Z",
  "supersedes": [],
  "expires_at": null,
  "formats": {
    "html": "https://feed7.dev/p/2607-22471v1-0kivszm",
    "json": "https://feed7.dev/p/2607-22471v1-0kivszm.json",
    "markdown": "https://feed7.dev/p/2607-22471v1-0kivszm.md"
  }
}