Gemini Web Client

Supports:

• Text generation
• Image generation (download + save)
• Reference images for vision input (attach local images)
• Multi-turn conversations via persisted

  --sessionId

Script Directory

Important: All scripts are located in the

scripts/

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:

scripts/main.ts

scripts/gemini-webapi/*

gemini_webapi

⚠️ Disclaimer (REQUIRED)

Before using this skill, the consent check MUST be performed.

Consent Check Flow

Step 1: Check consent file

# macOS
cat ~/Library/Application\ Support/baoyu-skills/gemini-web/consent.json 2>/dev/null

# Linux
cat ~/.local/share/baoyu-skills/gemini-web/consent.json 2>/dev/null

# Windows (PowerShell)
Get-Content "$env:APPDATA\baoyu-skills\gemini-web\consent.json" 2>$null

Step 2: If consent exists and

accepted: true

disclaimerVersion: "1.0"

Print warning and proceed:

⚠️  Warning: Using reverse-engineered Gemini Web API (not official). Accepted on: <acceptedAt date>

Step 3: If consent file doesn't exist or

disclaimerVersion

Display disclaimer and ask user:

⚠️  DISCLAIMER

This tool uses a reverse-engineered Gemini Web API, NOT an official Google API.

Risks:
- May break without notice if Google changes their API
- No official support or guarantees
- Use at your own risk

Do you accept these terms and wish to continue?

Use

AskUserQuestion

• Yes, I accept - Continue and save consent
• No, I decline - Exit immediately

Step 4: On acceptance, create consent file:

# macOS
mkdir -p ~/Library/Application\ Support/baoyu-skills/gemini-web
cat > ~/Library/Application\ Support/baoyu-skills/gemini-web/consent.json << 'EOF'
{
  "version": 1,
  "accepted": true,
  "acceptedAt": "<ISO timestamp>",
  "disclaimerVersion": "1.0"
}
EOF

# Linux
mkdir -p ~/.local/share/baoyu-skills/gemini-web
cat > ~/.local/share/baoyu-skills/gemini-web/consent.json << 'EOF'
{
  "version": 1,
  "accepted": true,
  "acceptedAt": "<ISO timestamp>",
  "disclaimerVersion": "1.0"
}
EOF

Step 5: On decline, output message and stop:

User declined the disclaimer. Exiting.

Quick start

npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello, Gemini"
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Explain quantum computing"
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

# Multi-turn conversation (agent generates unique sessionId)
npx -y bun ${SKILL_DIR}/scripts/main.ts "Remember this: 42" --sessionId my-unique-id-123
npx -y bun ${SKILL_DIR}/scripts/main.ts "What number?" --sessionId my-unique-id-123

Commands

Text generation

# Simple prompt (positional)
npx -y bun ${SKILL_DIR}/scripts/main.ts "Your prompt here"

# Explicit prompt flag
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Your prompt here"
npx -y bun ${SKILL_DIR}/scripts/main.ts -p "Your prompt here"

# With model selection
npx -y bun ${SKILL_DIR}/scripts/main.ts -p "Hello" -m gemini-2.5-pro

# Pipe from stdin
echo "Summarize this" | npx -y bun ${SKILL_DIR}/scripts/main.ts

Image generation

# Generate image with default path (./generated.png)
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A sunset over mountains" --image

# Generate image with custom path
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cute robot" --image robot.png

# Shorthand
npx -y bun ${SKILL_DIR}/scripts/main.ts "A dragon" --image=dragon.png

Vision input (reference images)

# Text + image -> text
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Describe this image" --reference a.png

# Text + image -> image
npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Generate a variation" --reference a.png --image out.png

Output formats

# Plain text (default)
npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello"

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

Options

--prompt <text>

-p

--promptfiles <files...>

--model <id>

-m

--image [path]

--reference <files...>

--ref <files...>

--sessionId <id>

--list-sessions

--json

--login

--cookie-path <path>

--profile-dir <path>

--help

-h

CLI note:

scripts/main.ts

--reference/--ref

--sessionId

Models

• gemini-3-pro

  - Default, latest model

• gemini-2.5-pro

  - Previous generation pro

• gemini-2.5-flash

  - Fast, lightweight

Authentication

First run opens a browser to authenticate with Google. Cookies are cached for subsequent runs.

Supported browsers (auto-detected in order):

• Google Chrome
• Google Chrome Canary / Beta
• Chromium
• Microsoft Edge

Override with

GEMINI_WEB_CHROME_PATH

# Force cookie refresh
npx -y bun ${SKILL_DIR}/scripts/main.ts --login

Environment variables

GEMINI_WEB_DATA_DIR

GEMINI_WEB_COOKIE_PATH

GEMINI_WEB_CHROME_PROFILE_DIR

GEMINI_WEB_CHROME_PATH

Proxy Configuration

If you need a proxy to access Google services (e.g., in China), set

HTTP_PROXY

HTTPS_PROXY

# Example with local proxy
HTTP_PROXY=http://127.0.0.1:7890 HTTPS_PROXY=http://127.0.0.1:7890 npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello"

# Image generation with proxy
HTTP_PROXY=http://127.0.0.1:7890 HTTPS_PROXY=http://127.0.0.1:7890 npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cat" --image cat.png

# Cookie refresh with proxy
HTTP_PROXY=http://127.0.0.1:7890 HTTPS_PROXY=http://127.0.0.1:7890 npx -y bun ${SKILL_DIR}/scripts/main.ts --login

Note: Environment variables must be set inline with the command. Shell profile settings (e.g.,

.bashrc

Examples

Generate text response

npx -y bun ${SKILL_DIR}/scripts/main.ts "What is the capital of France?"

Generate image

npx -y bun ${SKILL_DIR}/scripts/main.ts "A photorealistic image of a golden retriever puppy" --image puppy.png

Get JSON output for parsing

npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello" --json | jq '.text'

Generate image from prompt files

# Concatenate system.md + content.md as prompt
npx -y bun ${SKILL_DIR}/scripts/main.ts --promptfiles system.md content.md --image output.png

Multi-turn conversation

# Start a session with unique ID (agent generates this)
npx -y bun ${SKILL_DIR}/scripts/main.ts "You are a helpful math tutor." --sessionId task-abc123

# Continue the conversation (remembers context)
npx -y bun ${SKILL_DIR}/scripts/main.ts "What is 2+2?" --sessionId task-abc123
npx -y bun ${SKILL_DIR}/scripts/main.ts "Now multiply that by 10" --sessionId task-abc123

# List recent sessions (max 100, sorted by update time)
npx -y bun ${SKILL_DIR}/scripts/main.ts --list-sessions

Session files are stored in

~/Library/Application Support/baoyu-skills/gemini-web/sessions/<id>.json

• id

  : Session ID

• metadata

  : Gemini chat metadata for continuation

• messages

  : Array of

  {role, content, timestamp, error?}

• createdAt

  ,

  updatedAt

  : Timestamps

Extension Support

Custom configurations via EXTEND.md.

Check paths (priority order):

1. .baoyu-skills/baoyu-danger-gemini-web/EXTEND.md

   (project)

2. ~/.baoyu-skills/baoyu-danger-gemini-web/EXTEND.md

   (user)

If found, load before workflow. Extension content overrides defaults.