omni-ai-mcp 4.0.0

Multi-AI MCP bridge: Gemini + OpenRouter (400+ models) — text, code, image, video, TTS, RAG, Deep Research

omni-ai-mcp

20 AI tools for Claude Code — Gemini's unique capabilities (video, TTS, 1M context, RAG, Deep Research) plus 400+ models via OpenRouter. One MCP server, every AI.

---

What's New in v4.0.0

Multi-Provider: Gemini + OpenRouter

# Ask any model — Gemini or 400+ via OpenRouter
ask_model("Explain quantum computing", model="openai/gpt-4o")
ask_model("Write a poem", model="meta-llama/llama-3.3-70b")
ask_model("Review this code", model="gemini-3.1-pro-preview")  # auto-routes to Gemini

# Discover available models
gemini_list_models()

Dynamic Model Registry

No more hardcoded model IDs. The server discovers available models at runtime and always picks the best one. If a model is deprecated, it automatically falls back.

# config.py defaults are live-verified against the API on startup
# Override via env vars if needed:
export GEMINI_MODEL_PRO=gemini-3.1-pro-preview
export OPENROUTER_DEFAULT_MODEL=openai/gpt-4o

---

20 Tools

Multi-Provider (NEW)

Tool	Description
ask_model	Gemini or 400+ models via OpenRouter — auto-routes from model name
gemini_list_models	Live model discovery: Gemini + OpenRouter, deprecation warnings

Text & Reasoning

Tool	Description	Model
ask_gemini	Text generation with thinking mode, dual storage (local/cloud)	Gemini 3.1 Pro
gemini_code_review	Security, performance, quality analysis	Gemini 3.1 Pro
gemini_brainstorm	Creative ideation with 6 methodologies	Gemini 3.1 Pro
gemini_challenge	Devil's advocate — find flaws in ideas/plans/code	Gemini 3.1 Pro

Code

Tool	Description	Model
gemini_analyze_codebase	Whole-codebase analysis up to 1M tokens / 5MB	Gemini 3.1 Pro
gemini_generate_code	Structured code gen with dry-run preview	Gemini 3.1 Pro

Research & Web

Tool	Description	Model
gemini_web_search	Real-time search with Google grounding & citations	Gemini 2.5 Flash
gemini_deep_research	Autonomous 5–60 min research, 40+ sources	Deep Research Agent

RAG

Tool	Description
gemini_file_search	Query documents with citations
gemini_create_file_store	Create document stores
gemini_upload_file	Upload PDF, DOCX, code, etc.
gemini_list_file_stores	List available stores

Media (Gemini exclusive)

Tool	Description	Model
gemini_analyze_image	Vision: describe, OCR, Q&A	Gemini 2.5 Flash
gemini_generate_image	Imagen — up to 4K	Gemini 3.1 Pro Image
gemini_generate_video	Veo 3.1 — 4-8s with native audio	Veo 3.1
gemini_text_to_speech	30 voices, multi-speaker	Gemini 2.5 Flash TTS

Conversation

Tool	Description
gemini_list_conversations	List history: title, mode, turns, last activity
gemini_delete_conversation	Delete by ID or title

---

Quick Start

Prerequisites

• Python 3.9+
• Claude Code CLI
• Gemini API key — get one free
• (Optional) OpenRouter API key — openrouter.ai for 400+ models

Install

git clone https://github.com/marmyx77/omni-ai-mcp.git
cd omni-ai-mcp

# Gemini only
./setup.sh YOUR_GEMINI_API_KEY

# Gemini + OpenRouter (400+ models)
./setup.sh YOUR_GEMINI_API_KEY YOUR_OPENROUTER_KEY

Restart Claude Code. Verify:

claude mcp list
# omni-ai-mcp: Connected

Manual Install

pip install 'mcp[cli]>=1.0.0' 'google-genai>=1.55.0' pydantic defusedxml filelock

mkdir -p ~/.claude-mcp-servers/omni-ai-mcp
cp -r app/ run.py pyproject.toml ~/.claude-mcp-servers/omni-ai-mcp/

claude mcp add omni-ai-mcp --scope user \
  -e GEMINI_API_KEY=YOUR_KEY \
  -- python3 ~/.claude-mcp-servers/omni-ai-mcp/run.py

PyPI (coming soon)

pip install omni-ai-mcp
omni-ai-mcp-setup

---

Configuration

All settings via environment variables:

Variable	Default	Description
GEMINI_API_KEY	required	Google Gemini API key
OPENROUTER_API_KEY	—	OpenRouter key (enables ask_model for 400+ models)
GEMINI_MODEL_PRO	gemini-3.1-pro-preview	Override Pro model
GEMINI_MODEL_FLASH	gemini-2.5-flash	Override Flash model
GEMINI_MODEL_DEEP_RESEARCH	deep-research-pro-preview	Override research agent
OPENROUTER_DEFAULT_MODEL	openai/gpt-4o	Default OpenRouter model
GEMINI_SANDBOX_ROOT	cwd	Root for file access
GEMINI_SANDBOX_ENABLED	true	Enable path sandboxing
GEMINI_CONVERSATION_TTL_HOURS	3	Local conversation expiry
GEMINI_LOG_DIR	~/.omni-ai-mcp	Log & DB directory
GEMINI_LOG_FORMAT	text	json or text
GEMINI_DISABLED_TOOLS	—	Comma-separated tool names to disable

---

Examples

Ask any model

ask_model("Compare Python and Go for a REST API", model="openai/gpt-4o")
ask_model("Write unit tests for this function", model="gemini-3.1-pro-preview")
ask_model("Summarize this article")  # uses default model

Multi-turn conversations

ask_gemini("Analyze this architecture", mode="local", title="Arch Review")
# Returns: continuation_id: abc123

ask_gemini("What about security?", continuation_id="abc123")

# Cloud mode — 55-day retention, resume from any device
ask_gemini("Review my API", mode="cloud")
# Returns: continuation_id: int_abc123

Deep Research

gemini_deep_research("Latest AI agent frameworks comparison 2025", max_wait_minutes=30)
# Runs 5-30 min, returns structured report with 40+ citations

Codebase Analysis

gemini_analyze_codebase(
    prompt="Find security vulnerabilities",
    files=["src/**/*.py", "tests/*.py"],
    analysis_type="security"
)

Video Generation

gemini_generate_video(
    prompt="A time-lapse of a flower blooming in a sunlit field, 4K, cinematic",
    duration=8,
    resolution="1080p"
)

---

Architecture

omni-ai-mcp/
├── app/
│   ├── server.py              # FastMCP — 20 @mcp.tool() registrations
│   ├── core/                  # Config, logging, security
│   ├── services/
│   │   ├── gemini.py          # Gemini client + fallback logic
│   │   ├── model_registry.py  # Dynamic model discovery (NEW)
│   │   ├── openrouter.py      # OpenRouter client (NEW)
│   │   └── persistence.py     # SQLite conversation storage
│   ├── tools/                 # Tool implementations by domain
│   │   ├── text/              # ask_gemini, ask_model, models, etc.
│   │   ├── code/              # analyze_codebase, generate_code
│   │   ├── media/             # image, video, TTS
│   │   ├── web/               # web_search, deep_research
│   │   └── rag/               # file_store, file_search, upload
│   ├── schemas/               # Pydantic v2 validation
│   └── utils/                 # @file references, token estimation
├── tests/                     # 120+ tests (unit + integration)
├── .claude/
│   ├── commands/              # /gemini /gemini-research /gemini-review /gemini-models
│   └── agents/                # gemini-researcher, gemini-analyzer
├── setup.sh                   # One-command install
├── manifest.json              # .mcpb bundle manifest
└── pyproject.toml

---

Claude Code Plugin

Slash commands included in .claude/commands/:

Command	Action
/gemini <prompt>	Quick ask_gemini call
/gemini-research <topic>	Start deep research
/gemini-review <file>	Code review
/gemini-models	List available models
/ask-model [model] <prompt>	Ask any model (GPT-4o, Llama, Gemini, etc.)

Subagents in .claude/agents/ (auto-invoked by Claude Code):

Agent	Trigger	Capability
gemini-researcher	"research X", "find out about Y"	Deep Research Agent, 40+ sources
gemini-analyzer	"analyze codebase", "security audit"	1M token context window
model-orchestrator	"ask GPT-4o", "compare models", "use Llama"	Routes to 400+ models

---

Multi-Model Architecture

omni-ai-mcp uses Claude as the orchestrator with other models as tools. This is different from provider replacement:

User → Claude Code
           ↓ (orchestrates)
      omni-ai-mcp tools
      ├── ask_model("openai/gpt-4o")   → OpenRouter → GPT-4o
      ├── ask_model("meta-llama/...")  → OpenRouter → Llama 3
      ├── ask_gemini(...)              → Gemini API → Gemini Pro
      └── gemini_deep_research(...)    → Gemini API → Deep Research

Natural multi-model conversations

With the model-orchestrator subagent, Claude Code automatically routes to the right model:

"Ask GPT-4o to review this code"
→ model-orchestrator calls ask_model(model="openai/gpt-4o", ...)

"Compare how Gemini and Llama respond to this prompt"
→ model-orchestrator calls ask_model twice, presents comparison

"What's the best model for translating legal documents?"
→ model-orchestrator recommends and demonstrates

Alternative: Claude Code Router

For users who want to replace Claude's backend entirely (e.g., use Gemini as the primary model for all Claude Code interactions), claude-code-router is a complementary tool:

// ~/.claude-code-router/config.json
{
  "Providers": [{"name": "openrouter", "api_key": "sk-or-..."}],
  "Router": {
    "default": "openrouter,openai/gpt-4o",
    "think": "openrouter,anthropic/claude-3.5-sonnet"
  }
}

omni-ai-mcp = Claude orchestrates other models as tools claude-code-router = Replace Claude's backend with another model entirely

---

Security

• Path sandboxing: all file access restricted to GEMINI_SANDBOX_ROOT
• Secrets sanitization: API keys masked in logs
• XML sanitization: LLM output sanitized before parsing
• Atomic file writes with automatic backups
• Rate limiting via provider-side quotas

---

Development

# Run tests
python -m pytest tests/unit/ -v
python -m pytest tests/integration/ -v  # requires GEMINI_API_KEY

# Quick import check
python -c "from app.core.config import config; print(f'v{config.version}')"

# Reinstall after changes
cp -r app/ ~/.claude-mcp-servers/omni-ai-mcp/
cp run.py ~/.claude-mcp-servers/omni-ai-mcp/
# Restart Claude Code

---

License

MIT — see LICENSE