god-tibo-imagen（神提博图像）

god-tibo-imagen — AI Image Generation via Codex Backend

Zero dependencies. Reuses your Codex login. Works from CLI, Node.js, or Python.

god-tibo-imagen generates images by sending requests to Codex's ChatGPT backend. It reads existing local authentication from ~/.codex/auth.json — no separate API key setup required if you already use Codex CLI.

Installation

Plugin (Claude Code)

claude plugin marketplace add NomaDamas/god-tibo-imagen

npm (CLI — global)

npm install -g god-tibo-imagen
# Provides the `gti` command

npm (Library)

npm install god-tibo-imagen

Python SDK

pip install god-tibo-imagen

Skill (any platform)

npx skills add https://github.com/akillness/oh-my-skills --skill god-tibo-imagen

Requirements

• Node.js 20+ (CLI / library)
• Python 3.10+ (Python SDK)
• Existing Codex CLI login at ~/.codex/auth.json

When to use

• Generate AI images without a separate API key — reuses Codex auth
• Automate image generation from shell scripts, Node.js, or Python workflows
• Use reference images as context for guided generation (up to multiple images)
• Run dry-run validation before submitting real generation requests
• Integrate image generation into agent pipelines via the Python Client class

Do not use when

• You need a supported public API (this uses an undocumented internal backend — may change)
• You do not have an active Codex / ChatGPT subscription with a valid ~/.codex/auth.json
• You need guaranteed SLA or production-grade image generation → use DALL-E API or Stable Diffusion

CLI usage

# Basic generation
gti --prompt "blue square icon on white background" --output ./out.png

# With reference image(s)
gti --prompt "make it round" --input ./ref1.png --input ./ref2.jpg --output ./out.png

# Dry-run (validate without generating)
gti --prompt "sunset over mountains" --dry-run

# Debug output (redacts sensitive data)
gti --prompt "cartoon cat" --output ./cat.png --debug

Node.js library

import { createProvider, resolveConfig } from 'god-tibo-imagen'

const config = await resolveConfig()
const provider = createProvider(config)
const result = await provider.generate({ prompt: 'abstract art' })

Python SDK

from god_tibo_imagen import Client

client = Client()
client.generate_image(
    prompt="blue square icon",
    output="./out.png"
)

Supported formats

Operating rules

1. Verify ~/.codex/auth.json exists and is valid before running
2. Use --dry-run to validate prompt and config without consuming quota
3. Treat the backend as undocumented — avoid hard dependencies in production code
4. Use --debug to inspect request shape; sensitive tokens are automatically redacted
5. Pass multiple --input flags for multi-image reference context

References

• NomaDamas/god-tibo-imagen
• npm: god-tibo-imagen
• Docs

Source: NomaDamas/god-tibo-imagen — MIT License