humantonylee/best-codex/imagegen

Claude Code and agent skill by humantonylee/best-codex

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Use this skill whenever the user asks to generate, create, render, draw, or make an image, picture, illustration, icon, logo, or any other visual asset. Also use it when the user wants to edit, modify, restyle, or combine existing images (provide them as reference_file inputs). The skill drives the `imagegen` CLI in this repo, which uses the user's local Codex ChatGPT authentication to call the private image-generation backend. Trigger phrases include "generate an image", "create a picture", "make an image of", "draw me a", "render this", "make this cat wear a hat", and similar. Prefer this skill over describing images in text.

Install

npx vskill@latest install humantonylee/best-codex/imagegen

SKILL.md

imagegen

Generate images from text prompts (and optional reference images) by running the imagegen CLI shipped in this repository (dist/cli.js).

How to invoke

Run the CLI from the repo root. It is not published to npm; invoke it via bun run cli (or bun run cli). The TypeScript source must be compiled first with bun run build.

Basic generation
bash
bun run build
bun run cli --prompt "flat blue square icon" --output-dir ./outputs/generate
With reference images

Pass --reference_file <path> (or -r) one or more times to use existing images as input. Supported formats: png, jpg/jpeg, gif, webp.

bash
bun run cli -p "Make this cat wear a hat" -r ./cat.png -o ./outputs/generate --output-prefix cat-hat
Aspect ratio / size

Pass --aspect_ratio <ratio> (or -a). Accepts 1:1, 3:2, 2:3, 16:9, 9:16, 4:3, 3:4, auto, or a raw size like 1536x1024, 2048x1152, 3840x2160, 2160x3840.

bash
bun run cli -p "a sunset over mountains" -a 16:9 -o ./outputs/generate --output-prefix sunset
Number of images

Pass --number <int> (or -n) to generate up to 4 images in one call (private-codex provider only).

bash
bun run cli -p "a friendly robot" -n 2 -o ./outputs/generate --output-prefix robot
Provider selection

Pass --provider <name>:

  • private-codex (default): direct HTTP to the private Codex backend. Fast,

supports --number, --aspect_ratio, and --reference_file.

  • codex-cli: headless codex exec fallback. No reference images, no size

selection. Useful when the private HTTP path is rate-limited or down.

  • auto: try private-codex first, fall back to codex-cli on failure.

Fallback is refused when --number > 1, --aspect_ratio/size, or --reference_file are set (codex-cli cannot honor those, so we fail fast rather than silently drop them).

bash
bun run cli --provider codex-cli -p "a red square" -o ./outputs/generate --output-prefix red
Dry run

Validate auth and print the request shape without calling the backend.

bash
bun run cli -p "flat blue square icon" --dry-run

Required arguments

  • --prompt <text> (or -p) — required text prompt
  • --output-dir <dir> (or -o) — directory to save images (default: cwd)

Prerequisites the agent must check

All providers require the codex CLI authenticated via ChatGPT subscription OAuth. This is non-negotiable — no API keys, no service accounts, no fabricated auth.

  1. The codex binary must be on PATH. If missing, stop and tell the user

to install Codex — do not try to install it for them.

  1. The codex CLI must be logged in via ChatGPT subscription OAuth. Verify

with codex login status (must show "Logged in using ChatGPT"). If it shows API key auth or is not logged in, stop and tell the user to run codex login and choose the ChatGPT option.

  1. Confirm ~/.codex/auth.json exists with auth_mode = chatgpt. Verify

with bun run cli --auth. If missing or wrong auth_mode, stop and tell the user — do not fabricate auth state.

These checks apply regardless of which --provider is selected. The private-codex provider reuses the same ChatGPT OAuth token from ~/.codex/auth.json; the codex-cli provider drives codex exec directly. Both depend on a valid ChatGPT subscription OAuth session.

After running

imagegen saves PNG(s) to --output-dir and prints a JSON summary that includes savedPath for each image. Report savedPath back to the user.

Hook script

A convenience hook script is provided at .agents/skills/imagegen/scripts/imagegen-hook.sh that wraps the CLI with sensible defaults and streams the JSON result to stdout. It is intended to be called from other tools or shell pipelines.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/humantonylee/best-codex/imagegen/badge)](https://verified-skill.com/skills/humantonylee/best-codex/imagegen)

Something wrong with this skill or its page? Tell us.