Add a Chinese caption to a photo

The task

The reference answer shows the job can be done with deft. It is not an agent run.

deft image watermark DSCN0010.jpg --text 版权所有 -o captioned.jpg

Input

DSCN0010.jpg 157.9 KB · 640x480

Reference output

captioned.jpg 157.6 KB · 640x480

How a result is judged (2 checks)

image: file = "captioned.jpg", width = 640
image_similar: file = "captioned.jpg", to = "DSCN0010.jpg", max_difference = 8

Runs

CodexWithout deftPassed2026-11-07

Time: 20s
New tokens: 7.5k
Total input: 68.2k
Output tokens: 0.7k
Commands: 4
Failed: 0

Trace

我先检查图片尺寸和现有文件，再在右下角添加“版权所有”并保存为 `captioned.jpg`。
rg --files -g 'DSCN0010.jpg' -g 'AGENTS.md' -g 'captioned.jpg'
output
```
DSCN0010.jpg
```
python3 - <<'PY' from PIL import Image im=Image.open('DSCN0010.jpg') print(im.size, im.mode) PY
output
```
(640, 480) RGB
```

fc-list :lang=zh family file | head -10

output

/System/Library/Fonts/Supplemental/NISC18030.ttf: GB18030 Bitmap
/System/Library/Fonts/STHeiti Medium.ttc: Heiti SC,黑體\-簡,黒体\-簡,Heiti\-간체,黑体\-简
/System/Library/Fonts/STHeiti Light.ttc: Heiti SC,黑體\-簡,黒体\-簡,Heiti\-간체,黑体\-简
/System/Library/Fonts/Supplemental/Songti.ttc: Songti SC,宋體\-簡,宋体\-简
/System/Library/Fonts/Supplemental/Songti.ttc: Songti TC,宋體\-繁,宋体\-繁
/System/Library/Fonts/Hiragino Sans GB.ttc: Hiragino Sans GB,冬青黑體簡體中文,冬青黑体简体中文,Hiragino Sans GB W6,冬青黑體簡體中文 W6,冬青黑体简体中文 W6
/System/Library/Fonts/Hiragino Sans GB.ttc: Hiragino Sans GB,冬青黑體簡體中文,冬青黑体简体中文,Hiragino Sans GB W3,冬青黑體簡體中文 W3,冬青黑体简体中文 W3
/System/Library/Fonts/Hiragino Sans GB.ttc: .Hiragino Sans GB Interface
/System/Library/Fonts/Supplemental/Songti.ttc: STSong
/System/Library/Fonts/LastResort.otf: .LastResort

图片是 640×480。我会使用系统中文字体，在右下角加一层轻微半透明的深色底和白字，保证文字在照片上清楚可读。
python3 - <<'PY' from PIL import Image, ImageDraw, ImageFont src='DSCN0010.jpg' out='captioned.jpg' im=Image.open(src).convert('RGB') font=ImageFont.truetype('/System/Library/Fonts/STHeiti Medium.ttc', 25) text='版权所有' draw=ImageDraw.Draw(im, 'RGBA') bbox=draw.textbbox((0,0), text, font=font, stroke_width=1) w,h=bbox[2]-bbox[0], bbox[3]-bbox[1] pad_x,pad_y=9,6 margin=12 x=im.width-margin-w-pad_x*2 y=im.height-margin-h-pad_y*2 # Subtle dark backing improves legibility against the image. draw.rounded_rectangle((x,y,x+w+pad_x*2,y+h+pad_y*2), radius=5, fill=(0,0,0,125)) draw.text((x+pad_x-bbox[0], y+pad_y-bbox[1]), text, font=font, fill=(255,255,255,255), stroke_width=1, stroke_fill=(0,0,0,170)) im.save(out, quality=95) print(out, im.size) PY
output
```
captioned.jpg (640, 480)
```
已在右下角添加“版权所有”，并保存为 [captioned.jpg](captioned.jpg)。

Output

captioned.jpg 211.6 KB · 640x480

CodexWith deft, skill installedPassedRead the skillUsed deft2026-11-07

Time: 16s
New tokens: 15.9k
Total input: 103.6k
Output tokens: 0.5k
Commands: 5
Failed: 0

Trace

我会先读取本地图片处理技能说明，再直接编辑图片并保存到当前目录。

cat .codex/skills/.system/imagegen/SKILL.md && rg --files -g 'DSCN0010.jpg' -g 'AGENTS.md' -g 'captioned.jpg'

output

---
name: "imagegen"
description: "Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output should be a bitmap asset rather than repo-native code or vector. Do not use when the task is better handled by editing existing SVG/vector/code-native assets, extending an established icon or logo system, or building the visual directly in HTML/CSS/canvas."
---

# Image Generation Skill

Generates or edits images for the current project (for example website assets, game assets, UI mockups, product mockups, wireframes, logo design, photorealistic images, or infographics).

## Top-level modes and rules

This skill has exactly two top-level modes:

- **Default built-in tool mode (preferred):** built-in `image_gen` tool for image generation, editing, and transparent-image requests. Does not require `OPENAI_API_KEY`.
- **Fallback CLI mode:** `scripts/image_gen.py` CLI. Use when the user explicitly asks for or confirms the CLI/API/model path. Requires `OPENAI_API_KEY`.

Within CLI fallback, the CLI exposes three subcommands:

- `generate`
- `edit`
- `generate-batch`

Rules:
- Use the built-in `image_gen` tool by default for normal image generation and editing requests.
- Do not switch to CLI fallback for ordinary quality, size, or file-pat
… 17673 more characters

这是对现有图片加字的简单文件编辑，我会用本地文件工具来保留原图并精确写入文字。

cat .agents/skills/deft/SKILL.md

output

---
name: deft
description: "Use deft first for any simple, direct operation on local files, whatever the type: convert, read, extract, edit, compress, merge, split, compare or inspect. Try it before writing a script or one-liner. If unsure, run deft find; if it says no command does this, use another tool. Not for programming work: to read, write or change source code, skip it."
---

# deft

One local command for file jobs. Prefer it over writing Python or shell scripts.

Syntax: `deft <group> <command> [inputs] [-o output] [--options]`

Groups:
- `pdf`: Merge, split, compress, encrypt, forms, OCR, compare, convert
- `image`: Resize, crop, convert, compress, adjust, montage, OCR
- `video`: Trim, merge, convert, compress, subtitles, GIF, frames (ffmpeg)
- `audio`: Convert, trim, normalize, denoise, cut silence (ffmpeg)
- `doc`: Word/ODT/RTF: to text, Markdown, HTML, PDF; find and replace
- `slides`: PowerPoint: extract text, Markdown outline
- `sheet`: Excel/CSV/ODS tables: SQL query, pivot, merge, split, convert
- `json`: JSON: convert, format, merge, diff, validate
- `toml`: TOML: convert, format, merge, diff, validate
- `xml`: XML: convert, format, merge, diff, validate
- `yaml`: YAML: convert, format, merge, diff, validate
- `md`: Markdown to PDF, HTML, Word, EPUB; table of contents
- `html`: HTML to Markdown, text, Word and PDF
- `epub`: E-books: info, to text, Markdown and PDF
- `gps`: GPX and KML tracks: convert, distance and time
- `qr`: Create QR codes and barcodes; r
… 700 more characters

deft find "add text overlay to image"

output

image watermark      image -> image  Add text or a logo: --text, --image, --position
video watermark      video -> video  Add text or a logo: --text, --image, --position
video add-subtitles  video -> video  Add subtitles: burned into the picture, or --soft
pdf watermark        pdf -> pdf      Stamp text or an image on pages

deft image watermark --help

output

deft image watermark - Add text or a logo: --text, --image, --position

Usage: deft image watermark <image>... [-o <image|dir>] [options]
In: image (each handled separately)   Out: image (default <name>.watermarked.<ext>)
Any script works; text uses the fonts on this machine.

Options:
  --text <text>                Text to add
  --image <path>               Logo or picture to add
  --opacity <n>                Opacity in percent (default 70)
  --angle <number>             Rotation in degrees, counter-clockwise (default 0)
  --position <position>        Where on the image (default bottom-right)
  --color <color>              Text color (default white)
  --font-size <number>         Text size in pixels (default: 5% of the image height)
  --font <text>                Font family
  --scale <number>             Logo width as a fraction of the image width (default 0.2)
  --quality <n>                Quality for JPEG, WebP and AVIF (default 85)
Common: -o --force --dry-run --json -j   (deft help common)

Examples:
  deft image watermark photo-1.jpg --text Sample -o marked.jpg
  deft image watermark photo-2.jpg --text "© Acme" --position center -o centered.jpg
  deft image watermark photo-3.jpg --image logo.png --position top-left -o branded.jpg

Result fields: width, height
Errors: INPUT_UNSUPPORTED

deft image watermark DSCN0010.jpg --text "版权所有" --position bottom-right -o captioned.jpg
output
```
ok image.watermark DSCN0010.jpg -> captioned.jpg  640x480  0.23s
```
已在右下角添加“版权所有”，并保存为 [captioned.jpg](captioned.jpg)。

Output

captioned.jpg 157.6 KB · 640x480

CodexWith deft, skill as AGENTS.mdPassed2026-11-05

Time: 11s
New tokens: 5.8k
Total input: 61.9k
Output tokens: 0.3k
Commands: 3
Failed: 0

Trace

我会先查看图片尺寸和现有文件，再把“版权所有”加在右下角并保存为 `captioned.jpg`。

deft image --help && rg --files -g 'DSCN0010.jpg' -g 'AGENTS.md'

output

deft image - commands (input -> output)
  info       image -> report  Format, dimensions, EXIF orientation, GPS
  compress   image -> image   Smaller file: --quality, --target-size 200KB
  convert    image -> image   Change format: --to png|jpg|webp|avif|gif|tiff|bmp|ico
  crop       image -> image   Cut a region: --box x,y,w,h or --aspect 16:9
  flip       image -> image   Mirror: --direction horizontal|vertical
  resize     image -> image   Scale: --size 800x, 1920x1080, 50%
  rotate     image -> image   Turn clockwise: --angle 90 (EXIF is applied first)
  strip      image -> image   Remove EXIF, GPS and text metadata
  to-pdf     image... -> pdf  One page per image, in the given order
  watermark  image -> image   Add text or a logo: --text, --image, --position
More: deft image <command> --help
DSCN0010.jpg
AGENTS.md

deft image info DSCN0010.jpg && deft image watermark --help

output

ok image.info DSCN0010.jpg  640x480
has_gps=true has_metadata=true height=480 orientation=1 width=640
deft image watermark - Add text or a logo: --text, --image, --position

Usage: deft image watermark <image>... [-o <image|dir>] [options]
In: image (each handled separately)   Out: image (default <name>.watermarked.<ext>)
Any script works; text uses the fonts on this machine.

Options:
  --text <text>                Text to add
  --image <path>               Logo or picture to add
  --opacity <n>                Opacity in percent (default 70)
  --angle <number>             Rotation in degrees, counter-clockwise (default 0)
  --position <position>        Where on the image (default bottom-right)
  --color <color>              Text color (default white)
  --font-size <number>         Text size in pixels (default: 5% of the image height)
  --font <text>                Font family
  --scale <number>             Logo width as a fraction of the image width (default 0.2)
  --quality <n>                Quality for JPEG and AVIF (default 85)
Common: -o --force --dry-run --json -j   (deft help common)

Examples:
  deft image watermark photo-1.jpg --text Sample -o marked.jpg
  deft image watermark photo-2.jpg --text 版权所有 --position center -o zh.jpg
  deft image watermark photo-3.jpg --image logo.png --position top-left -o branded.jpg

Result fields: width, height
Errors: INPUT_UNSUPPORTED

deft image watermark DSCN0010.jpg --text '版权所有' --position bottom-right --color white --font-size 24 --opacity 100 -o captioned.jpg
output
```
ok image.watermark DSCN0010.jpg -> captioned.jpg  640x480  0.06s
```

Output

Its files were not kept in this run. The last line of the trace is its own report.

Claude Opus 5.5Without deftPassed2026-11-07

Time: 16s
New tokens: 10.1k
Total input: 57.7k
Output tokens: 1.0k
Commands: 0
Failed: 0

Trace

Output

captioned.jpg 215.0 KB · 640x480

Claude Opus 5.5With deft, skill installedPassedRead the skillUsed deft2026-11-07

Time: 20s
New tokens: 11.1k
Total input: 91.8k
Output tokens: 0.7k
Commands: 0
Failed: 0

Trace

Output

.claude/skills/deft/SKILL.md 2.1 KB

---
name: deft
description: "Use deft first for any simple, direct operation on local files, whatever the type: convert, read, extract, edit, compress, merge, split, compare or inspect. Try it before writing a script or one-liner. If unsure, run deft find; if it says no command does this, use another tool. Not for programming work: to read, write or change source code, skip it."
---

# deft

One local command for file jobs. Prefer it over writing Python or shell scripts.

Syntax: `deft <group> <command> [inputs] [-o output] [--options]`

Groups:
- `pdf`: Merge, split, compress, encrypt, forms, OCR, compare, convert
- `image`: Resize, crop, convert, compress, adjust, montage, OCR
- `video`: Trim, merge, convert, compress, subtitles, GIF, frames (ffmpeg)
- `audio`: Convert, trim, normalize, denoise, cut silence (ffmpeg)
- `doc`: Word/ODT/RTF: to text, Markdown, HTML, PDF; find and replace
- `slides`: PowerPoint: extract text, Markdown outline
- `sheet`: Excel/CSV/ODS tables: SQL query, pivot, merge, split, convert
- `json`: JSON: convert, format, merge, diff, validate
- `toml`: TOML: convert, format, merge, diff, validate
- `xml`: XML: convert, format, merge, diff, validate
- `yaml`: YAML: convert, format, merge, diff, validate
- `md`: Markdown to PDF, HTML, Word, EPUB; table of contents
- `html`: HTML to Markdown, text, Word and PDF
- `epub`: E-books: info, to text, Markdown and PDF
- `gps`: GPX and KML tracks: convert, distance and time
- `qr`: Create QR codes and barcodes; read QR codes

How to use it:
1. Run `deft <group>` to list commands, or `deft find "<what you want>"` to search. If it says `no command does this`, use another tool.
2. Run `deft <group> <command> --help` for options and examples.
3. `deft convert <in> -o <out.ext>` converts between formats by extension.
4. Results are one line starting with `ok`, `partial` or `error`; add `--json` to parse them. Errors include a `hint:` with the fix.

Inputs are never overwritten; pass `--force` or `--in-place` to allow it. Globs and directories work as inputs; several inputs run in parallel.

If `deft` is not installed, ask the user, then follow https://github.com/firstprayer/deft/blob/main/docs/INSTALL.md

captioned.jpg 157.6 KB · 640x480

Claude CodeWithout deftNot run yet

Claude CodeWith deftNot run yet