Skip to content
Agentmart
Submit a tool

write-image-prompts

Write image-generator prompts and edit briefs from art direction, references, and asset needs.

Skill by JoseArron

IndexedMIT★ 2
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/josearron-write-image-prompts

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Write image-generator prompts and edit briefs from art direction, references, and asset needs. Use for images, visual series, style studies, or reusable prompt packs; writing prompts does not itself request generation.

Use cases: writing, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/write-image-prompts

What's inside

Skill name
write-image-prompts
Description
Write image-generator prompts and edit briefs from art direction, references, and asset needs. Use for images, visual series, style studies, or reusable prompt packs; writing prompts does not itself request generation.
License
MIT
Instructions
31 lines in SKILL.md · view on GitHub

Files (3)

  • PROMPT-PACK.md
  • SKILL.md
  • openai.yaml · agents/

SKILL.md (first lines)

# Write image prompts

Translate a visual brief into prompts that preserve identity while changing the requested subject, scene, or state.

## Read the visual rules

Choose subject, style, cultural context, and intended use from this brief. Use the selected art contract and approved anchors. Read only relevant parts. If direction is supplied in the request, work from it without requiring a document. For unresolved choices, offer a labeled candidate prompt; use `define-art-direction` only when defining direction is part of the task.

Inspect supplied references before describing or editing them. Distinguish style, identity, and composition references. Tell the generator which role each reference serves when its interface supports that control.

## Construct the prompt

Keep stable rules together and separate the variable request. [PROMPT-PACK.md](PROMPT-PACK.md) provides a portable structure for series and edits.

Describe the subject and action, framing and camera, shape language, contour, materials, palette roles, lighting, background, and output requirements that matter. Use concrete traits before broad labels. Include intended use and reading scale when they affect the result.

- For recurring subjects, preserve the defining features selected in the brief while varying only the requested attributes.
- For pitch or graphic assets, specify cutout boundaries, transparency when needed, empty space, cropping, and compatible visual weight. Put precise typography in editable layout code when that better serves the deliverable.
- Make each series prompt self-contained enough to paste into the target tool. "Same style" works only when the tool receives the reference or prior prompt.
- For an edit, state what changes and what must remain. Supply the target image through the supported reference mechanism.
- Use negative instructions for concrete likely failures. Avoid generic quality slogans and incompatible styles.

Default to portable natural-language prompts when the generator is unspecified. Adapt syntax, references, seeds, aspect ratios, and negative-prompt fields only to features supported by the named tool; verify unfamiliar controls. A repeated seed or phrase does not guarantee identity consistency.

## Deliver or generate within scope

If asked for prompts, return prompts. If asked for images, load the available image-generation skill and use its tool for raster work. For code-native or vector outputs, use `create-procedural-art` or the established vector system when appropriate.

Inspect generated results against the contract and anchors. Record visible drift and revise the smallest relevant prompt section. Do not call uninspected outputs consistent or production-ready. Retain prompt-to-output mappings and selected/rejected state for batches so `build-asset-library` can catalog the files.

From JoseArron/skills, MIT.

More from JoseArron/skills

Signals

Stars
2
Forks
0
Open issues
0
Last commit (this folder)
Oct 11, 2026 (18 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from JoseArron

Cite this page

write-image-prompts skill. Agent Mart. https://agentmart.dev/skills/josearron-write-image-prompts (last updated 2026-10-24)