agentleFS
Jewel591Jewel591/claude-with-codex/skills/codex-image-gen/SKILL.md

codex-image-gen

Skill1 starsChanged 4 months agoMITView on GitHub

When the task is to produce image assets (app icons, illustrations, placeholder art, draft marketing visuals, etc.), delegate to the tmux Codex pane instead of writing drawing scripts yourself. Applies to any task whose deliverable is one or more image files.

10 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use addskill, apply-firmware and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • ryanportfolioSkill

    addskill

    Create, import, update, or install repo or personal skills: runtime ownership, resources, discovery checks.

    Also in 6 projects that use this

  • ryanportfolioSkill

    apply-firmware

    Bring current folder (plain folder, existing repo, old or partial firmware copy) up to full Harness Firmware in place. Mirror external repo → adopt-repo.

    Also in 6 projects that use this

  • +10 more from ryanportfolio

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Locate/launch the Codex pane: use using-codex-collaboration's bundled codex-pane.sh (ensure/send/wait/capture — all its disciplines apply).

    line 18
  2. Distinctive:

    Compliance constraints: default to no third-party trademarks, logos, or brand-distinctive visual elements in the asset — nominative fair …

    line 22

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Codex Image Generation (delegation SOP)

Positioning & the exception to the hard line

using-codex-collaboration draws a hard line: "Codex reviews, never does." Image asset generation is a whitelisted exception to that line: an image is not a code implementation — it is a one-off deliverable, it never enters the code-review loop, and it does not amount to "Codex wrote feature code". The exception covers only the production of image files; code implementation, scripts that land in the repo, and feature development remain non-delegable, always.

Capability note

How Codex produces the image depends on its environment. If the pane has a native image-generation tool (e.g. an $imagegen-style capability in newer Codex versions), tell it to use that explicitly in the brief — don't leave the method to chance. If it has no native capability, it will typically write a throwaway drawing script — that's acceptable: the acceptance criterion is the output file, not the production method. Either way the script/tool residue must stay in a temp directory, never the repo.

Workflow

  1. Locate/launch the Codex pane: use using-codex-collaboration's bundled codex-pane.sh (ensure/send/wait/capture — all its disciplines apply). Image tasks aren't picky about the repo directory.
  2. Send the brief, which must include:
    • Purpose and content intent (one sentence, e.g. "app icon for X: dark background + brand-green database/pulse motif")
    • Hard constraints: exact dimensions (e.g. 1024×1024), format (PNG/JPG), alpha-channel requirements (App Store icons must have no alpha), color style
    • Compliance constraints: default to no third-party trademarks, logos, or brand-distinctive visual elements in the asset — nominative fair use covers naming another product in text, not reproducing its marks in your graphics; only deviate with an explicit license or after legal review
    • Absolute output path in a scratch directory you control — never a path inside the repo working tree. The "you are the sole writer to the working tree" guarantee applies to delegation too: Codex writes to scratch, and you copy the accepted file into the repo yourself.
  3. Poll and wait in the background (codex-pane.sh wait). On completion, view the generated image yourself with your image-reading tool — file-exists ≠ content-acceptable; verify dimensions (sips -g pixelWidth -g pixelHeight on macOS, or ImageMagick identify) and alpha.
  4. Iterate with concrete feedback when it falls short ("the column seam pokes through the top face" beats "doesn't look good"). The same pane accumulates context — no need to restate every constraint.
  5. Once the image is accepted, you copy it from scratch into the repo, and the engineering hookup (asset catalog, config, PR) stays with you — that's code, never delegated.

Known boundaries

  • Formal design work with high aesthetic demands (e.g. a final brand icon) should still come from a human designer — this skill covers "good-enough" assets (v1 icons, placeholders, diagrams).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.