{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-27-006",
    "source_story_id": "tmp-story-skillshift-web-agents",
    "edition_id": "mp-2026-10-27-morning-0109",
    "edition_url": "https://themachinepress.com/edition/2026-10-27",
    "position": 6,
    "story_type": "dispatch",
    "section": "developer-tools",
    "editorial_classification": "editorial",
    "headline": "The Skill Stayed the Same. The Website Did Not.",
    "slug": "the-skill-stayed-the-same-the-website-did-not",
    "dek": "SkillShift-Bench changes context, visual grounding and execution conditions across three self-hosted web systems; execution shifts hurt most.",
    "summary": "SkillShift-Bench changes context, visual grounding and execution conditions across three self-hosted web systems; execution shifts hurt most.",
    "body_text": "Reusable web-agent skills are usually tested in the environment that produced them. SkillShift-Bench instead applies controlled contextual, perceptual and execution changes across three self-hosted sites, measuring completion, interaction cost, skill use and failure. Every shift family degraded robustness, with execution changes placing the strongest pressure on learned skills and increasing both cost and failure. The authors frame reuse under change as a nonstationary local-executability problem and offer a test-time adaptation method that improves execution without rebuilding the entire skill library. These are controlled systems rather than the open web, but the benchmark isolates a practical deployment question: whether stored instructions remain actionable after the interface and available actions evolve.",
    "why_it_matters": "SkillShift-Bench changes context, visual grounding and execution conditions across three self-hosted web systems; execution shifts hurt most.",
    "limitations": [
      "These are controlled systems rather than the open web, but the benchmark isolates a practical deployment question: whether stored instructions remain actionable after the interface and available actions evolve."
    ],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-27-006/the-skill-stayed-the-same-the-website-did-not",
    "json_url": "https://themachinepress.com/story/mp-2026-10-27-006.json",
    "first_published_at": "2026-10-27T09:00:00.000-04:00",
    "modified_at": "2026-10-27T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-27-006-001",
        "text": "SkillShift-Bench changes context, visual grounding and execution conditions across three self-hosted web systems; execution shifts hurt most.",
        "source_ids": [
          "source-2026-10-27-006"
        ],
        "qualification": "These are controlled systems rather than the open web, but the benchmark isolates a practical deployment question: whether stored instructions remain actionable after the interface and available actions evolve."
      }
    ],
    "source_ids": [
      "source-2026-10-27-006"
    ],
    "tags": [
      "web agents",
      "skills",
      "robustness",
      "environment shift"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-27-006",
      "title": "arXiv preprint 2610.33528, original v1",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2610.33528v1",
      "canonical_url": "https://arxiv.org/abs/2610.33528v1",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-25T20:00:00.000-04:00",
      "accessed_at": "2026-10-27T08:18:12.114-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-27-006-001"
      ]
    }
  ],
  "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 Skill Stayed the Same. The Website Did Not.",
    "publisher": "The Machine Press",
    "published_at": "2026-10-27T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-27-006/the-skill-stayed-the-same-the-website-did-not"
  }
}