{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-25-007",
    "source_story_id": "tmp-story-specharness-agent-authority-boundary",
    "edition_id": "mp-2026-09-25-morning-0078",
    "edition_url": "https://themachinepress.com/edition/2026-09-25",
    "position": 7,
    "story_type": "dispatch",
    "section": "developer-tools",
    "editorial_classification": "editorial",
    "headline": "The Agent’s Completion Claim Beat Its Pass Rate by Up to 37.9 Points",
    "slug": "the-agent-s-completion-claim-beat-its-pass-rate-by-up-to-37-9-points",
    "dek": "SpecHarness separates an agent's proposal from evidence that requirements were actually satisfied.",
    "summary": "SpecHarness separates an agent's proposal from evidence that requirements were actually satisfied.",
    "body_text": "Researchers extracted 509 source-grounded directions from SkillsBench and tested seven models. They report that 79.6% to 86.4% of directions were satisfied, while agents' completion-claim rates exceeded official evaluator pass rates by 28.7 to 37.9 percentage points. SpecHarness compiles visible specifications into source-linked obligations and lets only qualified evidence establish governed state, leaving ambiguous requirements advisory. The work argues for an authority boundary, not merely another instruction prompt.",
    "why_it_matters": "SpecHarness separates an agent's proposal from evidence that requirements were actually satisfied.",
    "limitations": [
      "SpecHarness compiles visible specifications into source-linked obligations and lets only qualified evidence establish governed state, leaving ambiguous requirements advisory.",
      "The work argues for an authority boundary, not merely another instruction prompt."
    ],
    "importance": 9,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-25-007/the-agent-s-completion-claim-beat-its-pass-rate-by-up-to-37-9-points",
    "json_url": "https://themachinepress.com/story/mp-2026-09-25-007.json",
    "first_published_at": "2026-09-25T09:00:00.000-04:00",
    "modified_at": "2026-09-25T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-25-007-001",
        "text": "SpecHarness separates an agent's proposal from evidence that requirements were actually satisfied.",
        "source_ids": [
          "source-2026-09-25-007"
        ],
        "qualification": "SpecHarness compiles visible specifications into source-linked obligations and lets only qualified evidence establish governed state, leaving ambiguous requirements advisory."
      }
    ],
    "source_ids": [
      "source-2026-09-25-007"
    ],
    "tags": [
      "AI agents",
      "specifications",
      "verification"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-25-007",
      "title": "arXiv preprint 2609.29921",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.29921",
      "canonical_url": "https://arxiv.org/abs/2609.29921",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-24T10:55:14.000-04:00",
      "accessed_at": "2026-09-25T08:29:03.434-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-25-007-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "The Agent’s Completion Claim Beat Its Pass Rate by Up to 37.9 Points",
    "publisher": "The Machine Press",
    "published_at": "2026-09-25T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-25-007/the-agent-s-completion-claim-beat-its-pass-rate-by-up-to-37-9-points"
  }
}