Getting Started

Using ReadyPixl with AI agents (Claude Code, Codex and others)

Let Claude Code, Codex or any AI agent that can run terminal commands edit your images with ReadyPixl: install the CLI skill, build presets, and get reliable results.

Markdown All docs for AI

If you use an AI agent that can run commands on your computer — Claude Code, Codex, Cursor, Gemini CLI or similar — it can run ReadyPixl for you. You say "clean up every image in this folder with my ChatGPT cleanup preset," and the agent does it.

This works through the ReadyPixl CLI, a command-line tool installed with the desktop app. The agent runs the CLI; you never have to open a terminal yourself.

What you need

The ReadyPixl desktop app (Windows or macOS), installed and signed in. The 7-day free trial works too.
An AI agent that can run terminal commands.

The web version and Chrome extension don't have a CLI — they're for you to use directly. Anything you save there (like presets) still reaches the agent, because it syncs to the desktop app.

Set it up once (2 minutes)

Open the desktop app and go to Automation → CLI / Automation.
Click Export SKILL.md and choose your agent's skills folder. ReadyPixl creates a readypixl-skill folder with a SKILL.md file inside.

Claude Code: your project's .claude/skills/ folder, or ~/.claude/skills/ for every project.
Codex and other agents: wherever that agent keeps skills, or just tell the agent where the file is.

Open a new terminal (terminals opened before installing don't see the readypixl command).
Ask your agent: "Run `readypixl --version` to check ReadyPixl works."

That's it. The skill file teaches the agent every ReadyPixl command, every tool, every setting, and what each error means.

The skill keeps itself up to date. The desktop app updates on its own, so a saved skill file can go stale. The skill tells the agent to compare itself with readypixl docs (the live documentation built into the installed app) and refresh itself when they differ. You can also click Export SKILL.md again any time.

Presets are the secret to reliable results

An agent can set every tool and every setting by itself. But the more settings it has to choose, the more chances it has to get one wrong — especially smaller or cheaper models.

The reliable way: you build the pipeline in the editor, where you can see the preview, and save it as a preset. The agent only has to pick the preset name and the folder. The settings are exactly what you tested.

Build the pipeline in ReadyPixl (web, extension or desktop) and check the preview on a few images.
Save Current Preset with a clear name — e.g. ChatGPT cleanup, Square product photos, Web 1600 compressed.
Tell your agent: "Use the ReadyPixl skill with the ChatGPT cleanup preset on `D:\Images\new`."

The agent runs something like:

readypixl process "D:\Images\new" --preset "ChatGPT cleanup"

Presets you save anywhere reach the agent. Signed in with the same account, presets from the web version and the Chrome extension sync to the desktop app, and the CLI uses them. Build a preset in the browser at lunch, and your agent can use it on your desktop that evening. Ask the agent to run readypixl presets list to see them all.

Multi-step pipelines come from the editor. The agent can create simple one-tool presets on its own (readypixl presets create), but anything with several steps — remove background, trim, upscale, reposition — is best built and saved in the editor.

Things you can ask your agent

"List my ReadyPixl presets."
"Run my Square product photos preset on every folder inside D:\Shop\Incoming."
"Compress everything in this folder to high quality."
"Run Color Removal V2 on High directly on this one image, no preset."
"What settings does the Reposition tool have?" (the agent runs readypixl plugins show reposition)
"Set ReadyPixl to put results in a readypixl-output subfolder instead of next to the originals."
"Assign my ChatGPT cleanup preset to right-click slot A."
"Every morning, process whatever landed in my Downloads\readypixl folder." (the agent can write a scheduled script)

Where the results go

By default ReadyPixl saves a new file next to each original: image.png → image-readypixl.png. Your originals are never touched unless you ask for that. The agent can change this per run (--output <folder>, --overwrite) or permanently (a readypixl-output subfolder, or overwrite).

Paid AI tools and credits

AI Background Removal and AI Upscale work from the CLI too, using the credits on your account (15–20 credits per background removal, 10 per upscale). If you're out of credits, the agent gets a clear error and can tell you. The free Upscaler in the desktop app costs nothing.

When the agent says ReadyPixl needs you

The CLI reports problems with exit codes the skill already explains to the agent:

Code	What it means	What you do
3	Preset not found	Check the preset name — the agent can run `readypixl presets list`
4	Tool unavailable or not enough credits	Buy credits on your account page
5	Not signed in	Open the desktop app and sign in
6	Desktop license needs checking	Open the desktop app while online (or your trial ended)

Most of the time you won't see these: the CLI quietly asks the desktop app to refresh your sign-in and license in the background.

Agents can read our docs and feedback board too

This help center: every article is plain markdown for agents. Add .md to any article URL, or read the whole help center at once from helpcenter.readypixl.com/llms-full.txt.
The feedback board: feedback.readypixl.com/llms.txt tells agents how to read requests and answers on the feedback board.

So if your agent gets stuck, "check the ReadyPixl help center" is a useful thing to say.