{
  "schema_version": 2,
  "kind": "shortening-method",
  "format": "agent-skill",
  "id": "image-caption",
  "name": "Image caption",
  "category": "Writing",
  "summary": "Turn supplied image context into a short caption with accurate attribution.",
  "use_cases": [
    "Editorial captions",
    "Report image captions",
    "Collection labels"
  ],
  "word_count": 149,
  "url": "https://sho.rten.it/methods/image-caption/",
  "instructions_url": "https://sho.rten.it/methods/image-caption/SKILL.md",
  "skill_url": "https://sho.rten.it/methods/image-caption/SKILL.md",
  "json_url": "https://sho.rten.it/methods/image-caption/llms.txt",
  "plain_text_url": "https://sho.rten.it/methods/image-caption/prompt.txt",
  "license": "MIT",
  "sources_url": "https://sho.rten.it/sources/#image-caption",
  "skill_name": "image-caption",
  "skill_description": "Turn supplied image context into a short caption with accurate attribution. Use for Editorial captions, Report image captions, Collection labels.",
  "agents_md_url": "https://sho.rten.it/methods/image-caption/AGENTS.md",
  "sources": [],
  "instructions": "Write a short visible caption for the supplied image description and verified context. Use it to connect the image to the surrounding story or document. This is a caption, not a complete accessibility description.\n\nLead with the image's relevant subject or event. Add a supplied date, location, identification, or explanation only when it helps the reader understand why the image is included. Preserve required photographer credits and uncertainty about identification or timing. Clearly distinguish facts supplied as context from what the image alone shows.\n\nCut cataloging prose, repeated article text, obvious visual detail, and decorative adjectives. Do not invent names, emotions, motives, causes, dates, or unseen activity. Do not identify a person or location from appearance alone. If only a text description is available, use only its facts. Prefer one sentence plus a credit when needed. Return only the caption; do not add an accessibility claim or editing commentary.",
  "example": {
    "context": "Caption written from a supplied description and verified context",
    "before": "Image description: two volunteers stand beside stacks of timber in a workshop. Verified context: these are volunteers for the River Shed project, sorting reclaimed boards for new community-garden beds. The photograph was taken in the River Shed workshop in May 2026. The volunteers' names are not supplied. It is not known whether all the stacked timber will be used for the beds. Required credit: Photo: Jo Malik.",
    "after": "River Shed volunteers sort reclaimed boards for community-garden beds in the project's workshop, May 2026. Photo: Jo Malik.",
    "must_preserve": [
      "River Shed volunteers and workshop",
      "Reclaimed boards for community-garden beds",
      "May 2026",
      "Required photo credit",
      "No claim that all visible timber will be used"
    ],
    "omitted": [
      "Removes literal stack placement and missing-identification notes; makes no claim about unnamed people or all timber."
    ],
    "kind": "illustrative"
  }
}