edit-image — Claude Code skill: what it does, how to install
edit-image modifies existing picture files with OpenAI gpt-image-2 via Codex CLI, altering isolated details, applying brand colors, or generating matching variants.
MIT ★ 195 updated 2026-08-10
/plugin marketplace add Sateezg/codex-bridge
/plugin install codex-bridge@codex-bridgeWhat edit-image does
This tool forwards source assets to the codex-imagegen wrapper using --ref flags. The model keeps unmentioned elements unchanged and writes the output to a fresh file path. You can pass up to 4 reference images to split subject and style targets within the prompt.
Execution requires a Bash timeout of 300000 ms or higher. When transparent backgrounds are needed, the tool places the subject on a #00FF00 canvas and strips it using the bundled remove_chroma_key.py script. Native alpha channels require an OPENAI_API_KEY.
Every edit consumes ChatGPT plan quota. Standard pixel operations like rotating, resizing, or cropping are best handled directly by ImageMagick or sips without model inference.
What it helps with
- swap backgrounds while preserving the primary subject
- recolor interface icons to match hex brand palettes
- create new sibling assets aligned with reference design styles
- update labels and text on existing artwork
Example prompts for the agent
- Change the sky to sunset orange in
./assets/hero.pngand save to./out/hero-sunset.png, keep people identical - Recolor
./assets/icon.pngto#2563EBon white while keeping exact shapes - Make a settings gear icon matching the stroke and palette of
./assets/home.png
How to install edit-image
1Plugin marketplace
/plugin marketplace add Sateezg/codex-bridge/plugin install codex-bridge@codex-bridgeThis installs the whole codex-bridge plugin with all its skills
2Manually: put the files in place
~/.claude/skills/edit-image/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/edit-image/
Manually: put the files in place
.agents/skills/edit-image/SKILL.mdFor all projects use ~/.agents/skills/
Manually: put the files in place
.cursor/skills/edit-image/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install Sateezg/codex-bridge skills/edit-image2Manually: put the files in place
.github/skills/edit-image/SKILL.mdPlatform CLI
gemini skills install https://github.com/Sateezg/codex-bridge --path skills/edit-imageManually: put the files in place
.devin/skills/edit-image/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about edit-image
How to install edit-image in Claude Code?
Register the marketplace and install the package with /plugin marketplace add Sateezg/codex-bridge, followed by /plugin install codex-bridge@codex-bridge.
What does edit-image do?
It applies targeted changes to graphics via OpenAI gpt-image-2 using the Codex CLI, preserving unmentioned areas and outputting to a new path.
Can edit-image overwrite the original image?
The source file is never modified. The CLI wrapper always saves results into a separate destination and prints that path.
Source text
SKILL.md · MIT
---
description: Edit, restyle, or vary an EXISTING image file using OpenAI gpt-image-2 through the Codex CLI — change one thing and keep the rest, swap a background, recolor to a brand palette, add or fix text, or produce style-matched variants of a reference image. Use whenever the user points at an image file and asks to change, fix, restyle, clean up, or make a version of it.
---
# Edit an existing image (gpt-image-2 via Codex CLI)
Same wrapper as generation, plus `--ref` to attach the source image:
```bash
codex-imagegen "<what to change>" <output.png> --ref <source.png> [--size WxH]
```
The wrapper attaches the reference with `codex exec -i`, tells Codex to preserve
everything not mentioned, and prints the absolute path of the new file. **The
source file is never modified** — always write to a new path.
## Examples
```bash
# change one thing, keep the rest
codex-imagegen "change the sky to a sunset orange, keep the building and people identical" \
./out/hero-sunset.png --ref ./assets/hero.png
# recolor to a brand palette
codex-imagegen "recolor to a monochrome palette of #2563EB on white, keep the exact shapes" \
./out/icon-blue.png --ref ./assets/icon.png
# style-matched sibling asset
codex-imagegen "a settings gear icon in exactly the style of the reference: same stroke weight, same palette, same corner radius" \
./out/settings.png --ref ./assets/home.png
```
Up to four `--ref` images can be passed — useful when one is the subject and the
others are style references. Say in the prompt which is which.
## Writing the change description
Be surgical. The model preserves what you don't mention, so name the change and
then name what must survive it:
- Good: *"replace the text on the sign with 'OPEN 24/7', keep the font style, lighting and every other element identical"*
- Bad: *"make it better"* — produces an unrelated image
For a series of edits, chain them one at a time (each output becomes the next
`--ref`) rather than stacking five changes into one prompt.
## Rules
1. **Never overwrite the source.** Write to a new file; offer to replace only after
the user has seen the result. Note that Codex itself resists overwriting — the
wrapper prints the path actually written, which may be a versioned sibling like
`out-v2.png`. Use the printed path, not the one you asked for.
2. **Verify by viewing.** Read the output PNG with the Read tool and compare
against the request before reporting done. Regenerate with a sharper prompt if
the model drifted — at most 2 retries.
3. **Use a Bash timeout of at least 300000 ms** (5 minutes).
4. **For a transparent result**, regenerate the subject on a flat `#00FF00`
background (`#FF00FF` if the subject is green) and strip the key with the
helper Codex ships:
```bash
python "${CODEX_HOME:-$HOME/.codex}/skills/.system/imagegen/scripts/remove_chroma_key.py" \
--input ./keyed.png --out ./final.png --auto-key border --soft-matte \
--transparent-threshold 12 --opaque-threshold 220 --despill
```
Verify transparent corners and no colour fringe afterwards. True native
transparency needs Codex's CLI fallback plus an `OPENAI_API_KEY` — only raise
that option for hard subjects (hair, fur, glass, smoke) and let the user decide.
5. **Not for precise pixel work.** Cropping, resizing, rotating, format conversion
and compression are faster and lossless with ImageMagick or `sips` — use those
directly instead of regenerating.
6. Each edit spends the user's ChatGPT plan quota. Don't loop unprompted.
For multi-image jobs, hand the whole brief to the `codex-artist` subagent instead.
© Sateezg · MIT · Sateezg/codex-bridge