baoyu-danger-gemini-web

name	baoyu-danger-gemini-web
description	Generates images and text via reverse-engineered Gemini Web API. Supports text generation, image generation from prompts, reference images for vision input, and multi-turn conversations. Use when other skills need image generation backend, or when user requests "generate image with Gemini", "Gemini text generation", or needs vision-capable AI generation.

Gemini Web Client

Text/image generation via Gemini Web API. Supports reference images and multi-turn conversations.

Script Directory

Important: All scripts are located in the scripts/ subdirectory of this skill.

Agent Execution Instructions:

1. Determine this SKILL.md file's directory path as SKILL_DIR
2. Script path = ${SKILL_DIR}/scripts/<script-name>.ts
3. Replace all ${SKILL_DIR} in this document with the actual path

Script Reference:

Script	Purpose
scripts/main.ts	CLI entry point for text/image generation
scripts/gemini-webapi/*	TypeScript port of gemini_webapi (GeminiClient, types, utils)

Consent Check (REQUIRED)

Before first use, verify user consent for reverse-engineered API usage.

Consent file locations:

• macOS: ~/Library/Application Support/baoyu-skills/gemini-web/consent.json
• Linux: ~/.local/share/baoyu-skills/gemini-web/consent.json
• Windows: %APPDATA%\baoyu-skills\gemini-web\consent.json

Flow:

1. Check if consent file exists with accepted: true and disclaimerVersion: "1.0"
2. If valid consent exists → print warning with acceptedAt date, proceed
3. If no consent → show disclaimer, ask user via AskUserQuestion:
   • "Yes, I accept" → create consent file with ISO timestamp, proceed
   • "No, I decline" → output decline message, stop
4. Consent file format: {"version":1,"accepted":true,"acceptedAt":"<ISO>","disclaimerVersion":"1.0"}

---

Preferences (EXTEND.md)

Use Bash to check EXTEND.md existence (priority order):

# Check project-level first
test -f .baoyu-skills/baoyu-danger-gemini-web/EXTEND.md && echo "project"

# Then user-level (cross-platform: $HOME works on macOS/Linux/WSL)
test -f "$HOME/.baoyu-skills/baoyu-danger-gemini-web/EXTEND.md" && echo "user"

┌──────────────────────────────────────────────────────────┬───────────────────┐ │ Path │ Location │ ├──────────────────────────────────────────────────────────┼───────────────────┤ │ .baoyu-skills/baoyu-danger-gemini-web/EXTEND.md │ Project directory │ ├──────────────────────────────────────────────────────────┼───────────────────┤ │ $HOME/.baoyu-skills/baoyu-danger-gemini-web/EXTEND.md │ User home │ └──────────────────────────────────────────────────────────┴───────────────────┘

┌───────────┬───────────────────────────────────────────────────────────────────────────┐ │ Result │ Action │ ├───────────┼───────────────────────────────────────────────────────────────────────────┤ │ Found │ Read, parse, apply settings │ ├───────────┼───────────────────────────────────────────────────────────────────────────┤ │ Not found │ Use defaults │ └───────────┴───────────────────────────────────────────────────────────────────────────┘

EXTEND.md Supports: Default model | Proxy settings | Custom data directory

Usage

# Text generation
npx -y bun ${SKILL_DIR}/scripts/main.ts "Your prompt"
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Your prompt" --model gemini-2.5-pro

# Image generation
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cute cat" --image cat.png
npx -y bun ${SKILL_DIR}/scripts/main.ts --promptfiles system.md content.md --image out.png

# Vision input (reference images)
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Describe this" --reference image.png
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Create variation" --reference a.png --image out.png

# Multi-turn conversation
npx -y bun ${SKILL_DIR}/scripts/main.ts "Remember: 42" --sessionId session-abc
npx -y bun ${SKILL_DIR}/scripts/main.ts "What number?" --sessionId session-abc

# JSON output
npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello" --json

Options

Option	Description
--prompt, -p	Prompt text
--promptfiles	Read prompt from files (concatenated)
--model, -m	Model: gemini-3-pro (default), gemini-2.5-pro, gemini-2.5-flash
--image [path]	Generate image (default: generated.png)
--reference, --ref	Reference images for vision input
--sessionId	Session ID for multi-turn conversation
--list-sessions	List saved sessions
--json	Output as JSON
--login	Refresh cookies, then exit
--cookie-path	Custom cookie file path
--profile-dir	Chrome profile directory

Models

Model	Description
gemini-3-pro	Default, latest
gemini-2.5-pro	Previous pro
gemini-2.5-flash	Fast, lightweight

Authentication

First run opens browser for Google auth. Cookies cached automatically.

Supported browsers (auto-detected): Chrome, Chrome Canary/Beta, Chromium, Edge.

Force refresh: --login flag. Override browser: GEMINI_WEB_CHROME_PATH env var.

Environment 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
HTTP_PROXY, HTTPS_PROXY	Proxy for Google access (set inline with command)

Sessions

Session files stored in data directory under sessions/<id>.json.

Contains: id, metadata (Gemini chat state), messages array, timestamps.

Extension Support

Custom configurations via EXTEND.md. See Preferences section for paths and supported options.

🔒 Mandatory Post-Processing (EXIF strip)

After every image generated by this skill — regardless of file extension claimed (.png/.jpg/.webp), regardless of provider (Gemini/OpenAI/DashScope), regardless of whether the user is going to share it — Claude MUST run the post-processing pipeline defined in:

~/.baoyu-skills/image-post-processing.md

Why this is mandatory and unconditional:

• AI providers routinely embed metadata: provider tags, timestamps, color profiles, copyright fingerprints, watermark indicators, C2PA chunks
• Provider-claimed file format (e.g. .png) often disagrees with actual format (e.g. JPEG with PNG extension) — verify and rename
• Local-only files frequently get uploaded later (CDN, R2, screenshots into shared docs, git history) — strip on creation, not before sharing

Required steps for each generated image (in order):

1. Detect actual format with PIL.Image.open(f).format (not the extension)
2. Re-save via PIL with no metadata (this is the EXIF-strip step):

   from PIL import Image
   img = Image.open(path)
   data = list(img.getdata())
   clean = Image.new(img.mode, img.size)
   clean.putdata(data)
   clean.save(target_path, img.format, quality=95, optimize=True)
   

3. If actual format ≠ claimed extension, rename to the correct one and remove the original
4. Verify: re-open the saved file and assert img.getexif() is empty AND img.info has no metadata keys (other than benign jfif/dpi)
5. Only after verification passes may you report the image as done

⚠️ Skipping this step or trusting "the provider doesn't add metadata" is a contract violation. Verify, do not assume.