{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-22-026",
    "source_story_id": "tmp-story-access-policy-generation-tests",
    "edition_id": "mp-2026-09-22-morning-0075",
    "edition_url": "https://themachinepress.com/edition/2026-09-22",
    "position": 15,
    "story_type": "dispatch",
    "section": "developer-tools",
    "editorial_classification": "editorial",
    "headline": "Access Rules Worked Better When the Agent Had Tests",
    "slug": "access-rules-worked-better-when-the-agent-had-tests",
    "dek": "A structured Rego pipeline reached 50.3% strict correctness versus 15.3% for a direct prompt.",
    "summary": "A structured Rego pipeline reached 50.3% strict correctness versus 15.3% for a direct prompt.",
    "body_text": "Researchers translated natural-language access rules into Rego policies for Open Policy Agent using a pipeline that detects policy components, validates a schema, lints, compiles and generates positive and negative tests. On 372 annotated access-control statements, 50.3% of outputs passed the full correctness gate, versus 15.3% from a direct single-prompt baseline. The improvement is large, but roughly half still failed even under the structured approach. These are benchmark results, not a claim of production-ready access control.",
    "why_it_matters": "A structured Rego pipeline reached 50.3% strict correctness versus 15.3% for a direct prompt.",
    "limitations": [
      "These are benchmark results, not a claim of production-ready access control."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-22-026/access-rules-worked-better-when-the-agent-had-tests",
    "json_url": "https://themachinepress.com/story/mp-2026-09-22-026.json",
    "first_published_at": "2026-09-22T09:00:00.000-04:00",
    "modified_at": "2026-09-22T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-22-026-001",
        "text": "A structured Rego pipeline reached 50.3% strict correctness versus 15.3% for a direct prompt.",
        "source_ids": [
          "source-2026-09-22-015"
        ],
        "qualification": "These are benchmark results, not a claim of production-ready access control."
      }
    ],
    "source_ids": [
      "source-2026-09-22-015"
    ],
    "tags": [
      "access control",
      "policy as code",
      "testing"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-22-015",
      "title": "arXiv preprint 2609.24036",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.24036",
      "canonical_url": "https://arxiv.org/abs/2609.24036",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-20T23:02:03.000-04:00",
      "accessed_at": "2026-09-22T08:26:19.625-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-22-026-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "Access Rules Worked Better When the Agent Had Tests",
    "publisher": "The Machine Press",
    "published_at": "2026-09-22T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-22-026/access-rules-worked-better-when-the-agent-had-tests"
  }
}