Image generation skill using Gemini Web. Generates images from text prompts via Google Gemini. Also supports text generation. Use as the image generation backend for other skills like cover-image, xhs-images, article-illustrator.
Resources
1Install
npx skillscat add zsyggg/paper-craft-skills/baoyu-gemini-web Install via the SkillsCat registry.
This skill provides a programmatic interface to Google Gemini's web client for generating images and text from prompts. It addresses the need to use Gemini as an image generation backend within agent workflows, supporting reference image uploads, multi-turn conversations via session reuse, and experimental video generation. Developers should use it when building agent skills that require text-to-image generation or when integrating Gemini as the backend for cover-image, illustration, or content generation tasks.
Gemini Web Client
Supports:
- Text generation
- Image generation (download + save)
- Reference image upload (attach images for vision tasks)
- Multi-turn conversations within the same executor instance (
keepSession) - Experimental video generation (
generateVideo) — Gemini may return an async placeholder; download might require Gemini web UI
Quick start
npx -y bun scripts/main.ts "Hello, Gemini"
npx -y bun scripts/main.ts --prompt "Explain quantum computing"
npx -y bun scripts/main.ts --prompt "A cute cat" --image cat.png
npx -y bun scripts/main.ts --promptfiles system.md content.md --image out.png
# Multi-turn conversation (agent generates unique sessionId)
npx -y bun scripts/main.ts "Remember this: 42" --sessionId my-unique-id-123
npx -y bun scripts/main.ts "What number?" --sessionId my-unique-id-123Executor options (programmatic)
This skill is typically consumed via createGeminiWebExecutor(geminiOptions) (see scripts/executor.ts).
Key options in GeminiWebOptions:
referenceImages?: string | string[]Upload local images as references (vision input).keepSession?: booleanReuse GeminichatMetadatato continue the same conversation across calls (required if you want reference images to persist across multiple messages).generateVideo?: stringGenerate a video and (best-effort) download to the given path. Gemini may returnvideo_gen_chip(async); in that case you must open Gemini web UI to download the result.
Notes:
generateVideocannot be combined withgenerateImage/editImage.- When
keepSession=trueandreferenceImagesis set, reference images are uploaded once per executor instance.
Commands
Text generation
# Simple prompt (positional)
npx -y bun scripts/main.ts "Your prompt here"
# Explicit prompt flag
npx -y bun scripts/main.ts --prompt "Your prompt here"
npx -y bun scripts/main.ts -p "Your prompt here"
# With model selection
npx -y bun scripts/main.ts -p "Hello" -m gemini-2.5-pro
# Pipe from stdin
echo "Summarize this" | npx -y bun scripts/main.tsImage generation
# Generate image with default path (./generated.png)
npx -y bun scripts/main.ts --prompt "A sunset over mountains" --image
# Generate image with custom path
npx -y bun scripts/main.ts --prompt "A cute robot" --image robot.png
# Shorthand
npx -y bun scripts/main.ts "A dragon" --image=dragon.pngOutput formats
# Plain text (default)
npx -y bun scripts/main.ts "Hello"
# JSON output
npx -y bun scripts/main.ts "Hello" --jsonOptions
| Option | Description |
|---|---|
--prompt <text>, -p |
Prompt text |
--promptfiles <files...> |
Read prompt from files (concatenated in order) |
--model <id>, -m |
Model: gemini-3-pro (default), gemini-2.5-pro, gemini-2.5-flash |
--image [path] |
Generate image, save to path (default: generated.png) |
--sessionId <id> |
Session ID for multi-turn conversation (agent generates unique ID) |
--list-sessions |
List saved sessions (max 100, sorted by update time) |
--json |
Output as JSON |
--login |
Refresh cookies only, then exit |
--cookie-path <path> |
Custom cookie file path |
--profile-dir <path> |
Chrome profile directory |
--help, -h |
Show help |
CLI note: scripts/main.ts supports text generation, image generation, and multi-turn conversations via --sessionId. Reference images and video generation are exposed via the executor API.
Models
gemini-3-pro- Default, latest modelgemini-2.5-pro- Previous generation progemini-2.5-flash- Fast, lightweight
Authentication
First run opens Chrome to authenticate with Google. Cookies are cached for subsequent runs.
# Force cookie refresh
npx -y bun scripts/main.ts --loginEnvironment variables
| Variable | Description |
|---|---|
GEMINI_WEB_DATA_DIR |
Data directory |
GEMINI_WEB_COOKIE_PATH |
Cookie file path |
GEMINI_WEB_CHROME_PROFILE_DIR |
Chrome profile directory |
GEMINI_WEB_CHROME_PATH |
Chrome executable path |
Examples
Generate text response
npx -y bun scripts/main.ts "What is the capital of France?"Generate image
npx -y bun scripts/main.ts "A photorealistic image of a golden retriever puppy" --image puppy.pngGet JSON output for parsing
npx -y bun scripts/main.ts "Hello" --json | jq '.text'Generate image from prompt files
# Concatenate system.md + content.md as prompt
npx -y bun scripts/main.ts --promptfiles system.md content.md --image output.pngMulti-turn conversation
# Start a session with unique ID (agent generates this)
npx -y bun scripts/main.ts "You are a helpful math tutor." --sessionId task-abc123
# Continue the conversation (remembers context)
npx -y bun scripts/main.ts "What is 2+2?" --sessionId task-abc123
npx -y bun scripts/main.ts "Now multiply that by 10" --sessionId task-abc123
# List recent sessions (max 100, sorted by update time)
npx -y bun scripts/main.ts --list-sessionsSession files are stored in ~/Library/Application Support/baoyu-skills/gemini-web/sessions/<id>.json and contain:
id: Session IDmetadata: Gemini chat metadata for continuationmessages: Array of{role, content, timestamp, error?}createdAt,updatedAt: Timestamps