MultAI v0.2.260320A Alpha — Claude Cowork/Code Plugin

MultAIALPHA

Leverage the Knowledge and Intelligence of 7 Leading AIs at Once

A Claude Cowork/Code plugin that submits research prompts to Claude.ai, ChatGPT, Copilot, Perplexity, Grok, DeepSeek & Gemini in parallel — then synthesizes the results into structured reports, landscape analyses, and comparison matrices.

Get Started View on GitHub →

Supported Platforms

One Prompt. Seven Perspectives.

Each platform brings a different training set, reasoning style, and search index. MultAI queries all seven simultaneously via Playwright browser automation — no API keys needed.

Claude.ai

ChatGPT

Copilot

Perplexity

Grok

DeepSeek

Gemini

Platform	Free Tier	DEEP Mode	Agent Fallback
Claude.ai	✓	✓ Pro plan	✓
ChatGPT	✓	✓ Plus plan	✓
Microsoft Copilot	✓	—	✓
Perplexity	✓	✓ Pro plan	✓
Grok	✓	✓	✓
DeepSeek	✓	✓ DeepThink	✓
Google Gemini	✓	✓ Deep Research	✓

Capabilities

Everything You Need for Multi-AI Research

From parallel submission to structured reports — built into your Claude Code workflow.

⚙

Intelligent Routing

The orchestrator analyzes your intent and routes to the correct skill automatically — landscape, solution, comparator, or direct multi-AI.

⚡

Parallel Execution

All 7 platforms run concurrently via asyncio.gather(). A typical REGULAR-mode run completes in under 90 seconds.

📊

DEEP Research Mode

Activates each platform's research/think mode (Deep Research, DeepThink, etc.) for comprehensive answers with citations.

📋

Structured Reports

Consolidator synthesizes raw responses into structured deliverables — CIRs, 9-section landscape reports, or free-form summaries.

📈

XLSX Comparison Matrix

Maintain a capability matrix with weighted scores, combo columns, and auto-reordering. All via Python scripts.

🔒

Rate Limiting

Per-platform budgets, cooldowns, and daily caps — persisted across sessions. Never get rate-limited again.

🤖

Agent Fallback

When a Playwright selector fails, a vision-based browser-use agent takes over — powered by Claude Sonnet or Gemini Flash.

🛠

Self-Improving Skills

Each skill appends a timestamped run log after every execution and can update its own templates and scripts based on learnings.

🌍

Shared Domain Knowledge

domains/{domain}.md is enriched by both research skills after every run. New archetypes, terminology, and trends persist across sessions.

Skills

Five Skills, One Coherent Pipeline

Each skill handles a specific research workflow. The orchestrator routes automatically — or you can invoke directly.

/multai:orchestrator

Orchestrator

Intelligent router + engine owner. Entry point for all workflows.

Router Engine

Route decision — landscape, solution, comparator, or direct

Prompt confirmed / built

Pre-flight rate limit check + budget display

Engine invoked — parallel 7-platform submission

Results reviewed + consolidation

/multai:landscape-researcher

Landscape Researcher

Produces 9-section Market Landscape Reports for any solution category.

Research 9 Sections

Extract: category, audience, scope modifiers

Build prompt from template + domain knowledge

Engine run (DEEP mode, all platforms)

Consolidate into 9-section report

Launch report viewer + domain enrichment

/multai:solution-researcher

Solution Researcher

Deep-dive on a specific product — capabilities, pricing, competitive context.

Product CIR

Confirm product + scope + optional focus areas

Build research prompt from template

Engine run (DEEP mode) + consolidation into CIR

Optional: update comparison matrix

Domain knowledge enrichment

/multai:comparator

Comparator

Maintains XLSX capability matrices with weighted scores and combo columns.

XLSX Matrix

Parse operation: add, score, reorder, combo, verify

Read CIR or feature list

Execute via matrix_ops.py / matrix_builder.py

Validate: orphan check, score recalc, reorder

/multai:consolidator

Consolidator

Synthesizes raw multi-AI responses into unified structured reports. Called automatically by other skills.

Synthesis

Read raw archive + assess per-source reliability

Apply consolidation guide (if provided) as structural authority

Produce final report: consensus-first, union for 2+ sources

Workflow

How It Works

From natural language request to structured deliverable in four steps.

Describe Your Goal

"Research the DevOps platform landscape for SMBs"

Auto-Route

Orchestrator selects landscape-researcher and builds the prompt

7-AI Parallel Run

Playwright submits to all 7 platforms concurrently in DEEP mode

Structured Report

Consolidator produces a 9-section landscape report with Top 20 vendors

You (Claude Code) | v Orchestrator ── Phase 0: Route Decision | |--> landscape-researcher --> engine --> consolidator --> report |--> solution-researcher --> engine --> consolidator --> CIR --> comparator |--> comparator --> matrix_ops.py (direct) |--> direct multi-AI --> engine --> consolidator (generic)

Installation

Up and Running in 2 Commands

Python dependencies install automatically on your first session via a SessionStart hook. No manual setup.

# Step 1 — Register the marketplace (one-time) /plugin marketplace add alo-exp/multai # Step 2 — Install the plugin /plugin install multai@multai # Done! Start a new session and all deps install automatically.

# Or use Claude Desktop GUI: # Browse Plugins > Personal > + > Add marketplace from GitHub # Enter: alo-exp/multai # Then install the multai plugin # Or from the CLI: claude plugin marketplace add alo-exp/multai claude plugin install multai@multai

git clone https://github.com/alo-exp/multai.git cd multai bash install.sh # Load directly without marketplace registration: claude --plugin-dir ./multai

🔑

No API keys required for core functionality.
The engine uses Playwright browser automation with your existing Chrome logins. Optionally add GOOGLE_API_KEY (free) or ANTHROPIC_API_KEY for vision-based agent fallback.

Examples

Just Describe What You Need

The orchestrator routes to the right skill based on your intent. No slash commands needed.

Landscape Research

"Run a market landscape analysis of Internal Developer Platforms for platform engineering teams"

→ Routes to landscape-researcher
→ 9-section report + Top 20 vendors
        

Solution Research

"Research Humanitec.com — I want a Competitive Intelligence Report covering their IDP capabilities"

→ Routes to solution-researcher
→ CIR + optional XLSX update
        

Comparator

"Add Spinnaker to the comparison matrix based on the latest CIR"

→ Routes to comparator
→ Extracts features, ticks XLSX
        

Direct Multi-AI

"What are the main trade-offs between microservices and monoliths?"

→ Direct multi-AI run
→ 7 responses consolidated
        

Tier	Budget / Session	Cooldown	Daily Cap
free	3 requests	30 min	5
pro	10 requests	5 min	20

Architecture

Self-Contained Skills

Each skill owns its scripts, templates, and guides. No shared engine at the root — skills are portable modules.

multai/
├── skills/
│   ├── orchestrator/         # Router + Playwright engine
│   │   ├── SKILL.md
│   │   └── engine/              # orchestrator.py, platforms/, rate_limiter.py
│   ├── landscape-researcher/ # Market landscape reports
│   ├── solution-researcher/  # Product deep-dives (CIR)
│   ├── comparator/           # XLSX matrix ops scripts
│   └── consolidator/         # Response synthesis
├── domains/                  # Shared domain knowledge
├── reports/                  # Output + preview.html viewer
├── tests/                    # 96 pytest unit tests
└── docs/                     # SRS, Architecture, Test, CI/CD

MultAIALPHA

One Prompt. Seven Perspectives.

Everything You Need for Multi-AI Research

Intelligent Routing

Parallel Execution

DEEP Research Mode

Structured Reports

XLSX Comparison Matrix

Rate Limiting

Agent Fallback

Self-Improving Skills

Shared Domain Knowledge

Five Skills, One Coherent Pipeline

Orchestrator

Landscape Researcher

Solution Researcher

Comparator

Consolidator

How It Works

Describe Your Goal

Auto-Route

7-AI Parallel Run

Structured Report

Up and Running in 2 Commands

Just Describe What You Need

Built-In Budget Management

Comprehensive Docs Included

User Guide

Contributor Guide

Architecture

SRS

Test Strategy

CI/CD Strategy

Changelog

Self-Contained Skills