Skip to content

OpenSenseNova/SenseNova-Skills / skills/sn-image-imitate

sn-image-imitate

Verified

Generates a new image that imitates the style of a reference image while updating content based on user intent. Uses a three-stage pipeline: image annotation (long caption), caption rewriting, and image generation. Use when user asks to "imitate style", "保持这个风格重画", "按这张图风格生成", or "style transfer with new content".

FileKindSize
README.mdmarkdown5.7 KB
README_CN.mdmarkdown5.2 KB
SKILL.mdskill md15.1 KB
prompts/caption_rewrite.mdmarkdown1.2 KB
prompts/image_annotate.mdmarkdown3.3 KB
prompts/layout_review.mdmarkdown1.2 KB

prompts/image_annotate.md

View on GitHub
# Image Annotate Prompt

You are an expert in extracting and structuring content from images.

Complete the tasks below and output your response strictly in the specified format.

Step 1: Short Caption
Provide a concise short caption for the image in approximately 20 words, accurately describing its content.
Determine the primary language by estimating which language constitutes the majority of visible text, and use that language for your output; default to English if the image contains minimal or no legible text.

Step 2: Long Caption

Provide a detailed long caption describing the image comprehensively.
Enrich the description by including the title/subtitle, overall layout organization and style, chart type, data encoding methods, visual elements, and all textual content, enabling complete reconstruction of the image from the caption alone.
For the structural data in the image, include them into the caption in a structural format, e.g. table or multilevel list, following these guidelines:

- Extract all visible data without omitting secondary or partial information.
- For numeric values, both explicitly labeled values and those that must be actively inferred by reading axis scales, grid lines, data point positions, legends, or other visual encodings, record every derivable value exactly and completely, without omission, rounding, or approximation.
- For text values, preserve the original text from the image without translation or interpretation.

If the image contains several sub parts, describe each part thoroughly and clarify their relations.
Use natural, descriptive language while maintaining authenticity and accuracy; avoid generalizations. The caption length is unconstrained.
Determine the primary language by estimating which language constitutes the majority of visible text, and use that language for your output; default to English if the image contains minimal or no legible text.

Step 3: Layout Blueprint JSON

Provide a machine-readable layout blueprint to preserve composition when regenerating the image with new content.

Requirements:

- Output valid JSON object.
- Use normalized coordinates in `[0, 1]`.
- Keep values concise but complete enough for layout reconstruction.
- If information is uncertain, still provide best-estimate values and note uncertainty in `notes`.

Schema:

```json
{
  "canvas": {
    "aspect_ratio_hint": "16:9|9:16|1:1|other",
    "reading_flow": "left-to-right|top-to-bottom|radial|timeline|mixed"
  },
  "hierarchy": {
    "title_level_count": 0,
    "primary_focus_region_id": "region_1"
  },
  "regions": [
    {
      "id": "region_1",
      "role": "title|subtitle|legend|chart|kpi|text_block|image_block|icon_cluster|cta|other",
      "bbox_norm_xywh": [0.0, 0.0, 0.0, 0.0],
      "z_index": 0,
      "alignment": "left|center|right|justified|mixed",
      "content_density": "low|medium|high"
    }
  ],
  "relations": [
    {
      "from": "region_1",
      "to": "region_2",
      "type": "above|below|left_of|right_of|overlaps|contains|connected_to|sequence_next"
    }
  ],
  "style_lock": {
    "palette_mood": "",
    "typography_mood": "",
    "shape_language": "",
    "texture_rendering": "",
    "spacing_rhythm": ""
  },
  "notes": []
}
```

Output format:

SHORT_CAPTION:
`<short caption here>`

LONG_CAPTION:
`<long caption here>`

LAYOUT_BLUEPRINT_JSON:

```json
{ ... }
```
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall