chorus-ai 1.0.0

Query multiple AI platforms simultaneously

Chorus

Send one prompt to every major AI. See where they agree.

Chorus is a browser-native multi-AI consultation tool. Write one prompt, send it to ChatGPT, Claude, Gemini, Grok, Perplexity, DeepSeek, Mistral, and Copilot simultaneously — no API keys required. Uses your existing logged-in browser sessions and collects all responses into a D3 flowchart with consensus analysis, sentence-level diff, and Markdown export.

---

Why Chorus?

Other tools	Chorus
Require expensive API keys	Zero API keys — uses your browser sessions
Compare 2-3 AIs	8 AIs simultaneously
Raw text side-by-side	D3 tree + Consensus + Diff views
No account switching	Per-platform profile manager built in
Responses lost on refresh	Full history persisted to disk
Closed source / paid	Fully open source

---

Supported Platforms

Platform	URL used
🟢 ChatGPT	chatgpt.com
🟣 Claude	claude.ai/new
🌀 Gemini	gemini.google.com
✕ Grok	x.com/i/grok
🔭 Perplexity	perplexity.ai
🪟 Copilot	copilot.microsoft.com
🔵 DeepSeek	chat.deepseek.com
🔶 Mistral	chat.mistral.ai

---

Features

⚡ Zero API keys	Playwright drives your real browser — no tokens, no billing
🔀 Parallel querying	All AIs receive the prompt simultaneously via asyncio.gather
📡 Live progress	WebSocket streams each AI's status (queued → typing → done)
🌳 Tree view	D3.js radial tree — click any node to read the full response
📋 Cards view	Side-by-side response cards with copy button
🎯 Consensus view	Keyword agreement analysis — what all AIs agree on vs. split opinions
🔍 Diff view	Sentence-level diff — unique insights per AI highlighted in blue
🏷️ Prompt templates	5 built-in templates to get started quickly
📂 Persistent history	All sessions saved to disk — reload anytime
↓ Markdown export	Download any session as a formatted .md file
⚙️ Account switcher	Multiple accounts per platform, switch with one click
🧭 Onboarding wizard	First-run setup guides you through logging into each platform
🔁 Error recovery	Per-platform Retry and Re-login buttons — no full restart needed

---

Quick Start

# 1. Install
pip install chorus-ai

# 2. Install Playwright browser
playwright install chromium

# 3. Run
chorus

Open http://localhost:4747

First run: Chorus walks you through an onboarding wizard to log in to each platform. Sessions are saved in chorus/profiles/ and reused automatically.

---

Adding a New Account (e.g. work Google for Gemini)

1. Click the ⚙ icon next to any platform in the Send panel
2. Click + Add account and enter a name (e.g. work)
3. Chorus opens a fresh browser window — log in as your work account
4. Close the login window — the profile is saved
5. Select it from the dropdown next time you send

---

API

Method	Endpoint	Description
GET	/api/platforms	List platforms + metadata
GET	/api/platforms/{p}/profiles	List saved profiles
POST	/api/platforms/{p}/profiles/{name}	Create profile + open for login
POST	/api/query	Start a query session
POST	/api/sessions/{id}/retry/{platform}	Retry a failed platform
GET	/api/sessions/{id}	Session status + responses
GET	/api/sessions/{id}/consensus	Consensus analysis for a session
GET	/api/history	Saved session history
DELETE	/api/history/{id}	Remove a history item
GET	/api/export/{id}	Download session as Markdown
WS	/ws	Live platform status updates

---

Project Structure

chorus/
  pyproject.toml
  requirements.txt
  chorus/
    main.py                    # FastAPI app + session orchestration
    browser.py                 # Playwright BrowserManager (persistent profiles)
    websocket_manager.py       # WebSocket broadcast
    selectors.json             # All CSS selectors for every platform (one file)
    platforms/
      base.py                  # BaseAI ABC
      gemini.py
      chatgpt.py
      claude.py
      perplexity.py
      grok.py
      copilot.py
      deepseek.py
      mistral.py
  frontend/
    index.html                 # Full SPA (D3, WebSocket, consensus engine)
  profiles/                    # Browser profiles (gitignored)

---

Adding a New AI Platform

1. Create chorus/platforms/myai.py extending BaseAI
2. Implement submit_prompt() and wait_for_response()
3. Add selectors to chorus/selectors.json
4. Register in main.py PLATFORMS and PLATFORM_META dicts
5. That's it — the frontend picks it up automatically

---

Stack

• Backend: Python 3.10+, FastAPI, uvicorn
• Browser automation: Playwright (Chromium, persistent contexts)
• Frontend: Vanilla HTML/CSS/JS, D3.js v7
• Consensus: Server-side Jaccard similarity + TF keyword extraction

---

Contributing

See CONTRIBUTING.md. PRs welcome — especially new platform connectors, better selectors for existing platforms, and UI improvements.

License

MIT © Kabi10