10x-chat

10x-chat

Chat with web AI agents (ChatGPT, Gemini, Claude, Grok, Perplexity, NotebookLM) from your terminal via browser automation.

English | 繁體中文

10x-chat uses Playwright to automate browser sessions with persisted login profiles. Login once, then send prompts — bundled with file context — from your CLI or AI coding agent.

Use with OpenClaw

Paste this into your OpenClaw chat to install as a skill:

https://raw.githubusercontent.com/MikeChongCan/10x-chat/refs/heads/main/skills/10x-chat/SKILL.md

Quick Start

npx playwright install chromium  # one-time browser setup

# 1. Login to a provider (opens a browser window)
npx 10x-chat@latest login chatgpt

# 2. Send a prompt
npx 10x-chat@latest chat -p "Explain this error" --provider chatgpt --file "src/**/*.ts"

# 3. View session history
npx 10x-chat@latest status

[!TIP] Use bunx (bun.sh) instead of npx for faster startup.

Commands

login <provider>

Opens a headed browser for you to authenticate. The session persists across runs.

npx 10x-chat@latest login chatgpt       # Login to ChatGPT
npx 10x-chat@latest login gemini         # Login to Gemini
npx 10x-chat@latest login claude         # Login to Claude
npx 10x-chat@latest login grok           # Login to Grok
npx 10x-chat@latest login perplexity     # Login to Perplexity
npx 10x-chat@latest login notebooklm     # Login to NotebookLM
npx 10x-chat@latest login dreamina       # Login to Dreamina (CapCut) for video
npx 10x-chat@latest login --status       # Check login status for all providers

Google Flow (Veo) video uses your Google login — logging into Gemini covers it in shared-profile mode.

chat

Send a prompt to an AI provider via browser automation.

npx 10x-chat@latest chat -p "Review this code for bugs" --provider chatgpt --file "src/**/*.ts"
npx 10x-chat@latest chat -p "Debug this error" --file "logs/error.log"
npx 10x-chat@latest chat -p "Explain this" --dry-run              # Preview bundle without sending
npx 10x-chat@latest chat -p "Explain this" --copy                  # Copy bundle to clipboard
npx 10x-chat@latest chat -p "Long task" --timeout 600000 --headed  # 10min timeout, visible browser

delegate <provider>

Start a provider-specific JSONL delegate for external AI callers that need multi-step control instead of a single one-shot prompt. The delegate opens the provider once, keeps the tab alive, then reads JSON commands from stdin and writes JSON responses to stdout.

# One long-lived provider session controlled over JSONL
npx 10x-chat@latest delegate gemini --model Pro <<'EOF'
{"id":1,"action":"status"}
{"id":2,"action":"submit","prompt":"Draft three launch angles for this product."}
{"id":3,"action":"capture","timeoutMs":600000}
{"id":4,"action":"close"}
EOF

# Convenience combined action: select model/files, submit, capture
printf '%s\n' '{"action":"chat","model":"Thinking","prompt":"Review this pasted plan."}' \
  | npx 10x-chat@latest delegate gemini

Supported actions: status, goto, selectModel, attach, submit, capture, chat, eval, close.

image

Generate images via ChatGPT (DALL-E), Gemini (Imagen), or Grok with non-blocking polling.

npx 10x-chat@latest image -p "A fox astronaut in space, digital art" --provider chatgpt
npx 10x-chat@latest image -p "Watercolor landscape" --provider gemini --save-dir ./images
npx 10x-chat@latest image -p "Logo design" --provider grok --headed --timeout 120000

research

Deep research via ChatGPT, Gemini, or Perplexity with non-blocking progress polling. Designed for long-running research tasks (5-10+ minutes).

npx 10x-chat@latest research -p "Latest breakthroughs in quantum computing" --provider gemini
npx 10x-chat@latest research -p "Hard technical research" --provider gemini --model "Deep Think"
npx 10x-chat@latest research -p "Market analysis of EVs" --provider chatgpt --timeout 600000
npx 10x-chat@latest research -p "Compare React vs Vue in 2026" --provider perplexity --save-dir ./reports

video

Generate video via browser automation — Google Flow (Veo) or Dreamina (Seedance).

# Google Flow / Veo (default) — uses your Google login (shared with Gemini)
npx 10x-chat@latest video -p "A drone shot over snowy mountains at sunrise" --provider flow
npx 10x-chat@latest video -p "Neon city street, rain" --provider flow --model "Omni Flash" --duration 10 --orientation portrait
npx 10x-chat@latest video -p "She walks forward" --provider flow --image ref.png   # image-to-video

# Dreamina / Seedance (CapCut)
npx 10x-chat@latest login dreamina   # one-time CapCut login
npx 10x-chat@latest video -p "A paper boat in a rain gutter, macro" --provider dreamina --aspect 9:16 --dreamina-duration 4
npx 10x-chat@latest video -p "The glowing orb pulses and floats up" --provider dreamina --image ref.png --ref-mode omni

Shared flags:

Flow (Veo) — models: Omni Flash (default), Veo 3.1 - Lite, Veo 3.1 - Fast, Veo 3.1 - Quality.

Dreamina (Seedance) — requires login dreamina (CapCut account). Models: Seedance 2.0 Fast (default, cheapest), Seedance 2.0; Seedance 1.5 Pro / 1.0 / 1.0 Fast may be locked depending on your plan/region.

history

List chat history visible in provider sidebars (ChatGPT, Gemini, Claude, Grok, Perplexity).

npx 10x-chat@latest history --provider gemini
npx 10x-chat@latest history --provider all --limit 10
npx 10x-chat@latest history --provider chatgpt --json

status

List local 10x-chat CLI sessions.

npx 10x-chat@latest status              # Last 24 hours
npx 10x-chat@latest status --hours 72   # Last 3 days

session <id>

View details of a specific session.

npx 10x-chat@latest session <id> --render   # Pretty-print the response

config

View or modify configuration.

npx 10x-chat@latest config show
npx 10x-chat@latest config set provider gemini
npx 10x-chat@latest config set timeout 600000
npx 10x-chat@latest config set headless false

skill

Manage the agent integration skill (for Codex, Claude Code, etc).

npx 10x-chat@latest skill install   # Install SKILL.md to ~/.codex/skills/
npx 10x-chat@latest skill show      # Display SKILL.md content

notebooklm (alias: nb)

Manage NotebookLM notebooks and sources via RPC API.

npx 10x-chat@latest notebooklm list                              # List all notebooks
npx 10x-chat@latest notebooklm create "Research Topic"            # Create a notebook
npx 10x-chat@latest notebooklm delete <notebookId>                # Delete a notebook
npx 10x-chat@latest notebooklm sources <notebookId>               # List sources in notebook
npx 10x-chat@latest notebooklm add-url <notebookId> <url>         # Add URL source
npx 10x-chat@latest notebooklm add-url <notebookId> <url> --wait  # Add URL and wait for processing
npx 10x-chat@latest notebooklm add-file <notebookId> ./paper.pdf  # Upload file source
npx 10x-chat@latest notebooklm add-text <id> "Title" "Content"    # Add pasted text source
npx 10x-chat@latest notebooklm summarize <notebookId>             # AI summary + suggested topics

# Then chat with the notebook's sources:
npx 10x-chat@latest chat -p "Summarize key points" --provider notebooklm

migrate

Merge older per-provider isolated browser profiles into the single shared profile used by default. Useful when upgrading from a version that stored a separate profile per provider.

npx 10x-chat@latest migrate --dry-run            # Preview without making changes
npx 10x-chat@latest migrate                      # Auto-pick the largest profile as the base
npx 10x-chat@latest migrate --source chatgpt     # Use a specific provider's profile as the base
npx 10x-chat@latest migrate --keep               # Keep the old isolated profiles after migrating

File Bundling

The --file flag accepts globs. Files are assembled into a markdown bundle sent as the prompt:

npx 10x-chat@latest chat -p "Review these" --file "src/**/*.ts" "!src/**/*.test.ts"

Security-sensitive files (.env*, *.pem, *.key, etc.) are automatically excluded.

Data Layout

~/.10x-chat/
├── profiles/
│   ├── chatgpt/          # Playwright persistent browser profile
│   ├── gemini/
│   ├── claude/
│   ├── grok/
│   └── notebooklm/       # NotebookLM browser profile (shared Google auth)
├── sessions/
│   └── <uuid>/
│       ├── meta.json     # Session metadata
│       ├── bundle.md     # Prompt bundle sent
│       └── response.md   # Captured response
└── config.json           # User configuration

Agent Integration

10x-chat includes a SKILL.md for AI coding agents. Install it with:

npx 10x-chat@latest skill install

This lets agents like Codex or Claude Code use 10x-chat to query other models for cross-validation, code review, or debugging help.

Supported Providers

Development

bun install
bun run dev login chatgpt      # Run CLI in dev mode
bun run typecheck               # Type check
bun run lint                    # Lint
bun run test                    # Run tests
bun run build                   # Build for production

Publishing

Releases are automated via GitHub Actions. Push a version tag to publish:

npm version patch   # or minor / major
git push --follow-tags

Requires NPM_TOKEN secret in the GitHub repository settings.

Disclaimer

This project is provided for research and educational purposes only. It automates web interfaces that may be subject to the terms of service of the respective providers. You are solely responsible for how you use it, including compliance with any applicable terms, and you use it at your own risk.

License

MIT © Mike Chong — see LICENSE.