AI Integrations

Every GoOmni API endpoint works as a tool for AI assistants. Your customers' existing AI tools (Google Antigravity, ChatGPT, Claude, or any LLM with function calling) can discover and execute the whole GoOmni API automatically. No custom code needed.

Why this matters: You give your customers one API that their AI already knows how to use, so you build the integration once. Their agent reads the tool definitions, understands what each endpoint does, and calls them on behalf of the user.

Jump to platform

Google Antigravity Claude Desktop Cursor IDE ChatGPT Custom GPTs Google Gemini OpenAI API Anthropic API Any LLM

Authentication

AI clients connect to the MCP endpoint at /api/mcp using an access key, passed as a Bearer token. Generate one in Settings → MCP access. The MCP endpoint accepts the key and resolves it to your GoOmni session server-side, so the agent can call any tool on your behalf.

Generate a key, then pass it as a Bearer token

# 1. In GoOmni: Settings → MCP access → Generate access key
# 2. Every MCP client below uses it as:
#    Authorization: Bearer <your-mcp-access-key>

# Verify it works:
curl https://agent.goomni.ai/api/mcp \
  -H "Authorization: Bearer <your-mcp-access-key>" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Direct REST calls are different. If you are calling GoOmni endpoints directly (not through an MCP client), authenticate with a session cookie instead of a Bearer token. See the Authentication guide.

Google Antigravity

Recommended

Agent-first IDE powered by Gemini 3

Antigravity uses MCP (Model Context Protocol) to connect AI agents to external tools. Add GoOmni as an MCP server and the agent can search knowledge bases, generate test cases, analyze websites, and call any GoOmni endpoint, all from within the IDE.

Setup

In Antigravity, click ... → MCP Servers → Manage MCP Servers → View raw config
Add this to your mcp_config.json:

mcp_config.json

{
  "mcpServers": {
    "goomni": {
      "url": "https://agent.goomni.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer <your-mcp-access-key>"
      }
    }
  }
}

Click Refresh in the MCP Servers panel
Antigravity discovers all GoOmni tools, and the agent can call them directly

Example prompts

"Search my knowledge base for ADA accessibility requirements"
"Generate QA test cases for project proj_abc123"
"Crawl example.com and run a Lighthouse SEO audit"
"Extract requirements from the uploaded RFP document"

Claude Desktop

Anthropic's desktop app with MCP support

Claude Desktop loads only command-based (stdio) MCP servers from its config file, so the url form used by other clients is skipped. Use the one-click bundle, or bridge the hosted endpoint with the mcp-remote proxy.

Option A — One-click bundle (recommended)

In GoOmni, open Settings → MCP access and click Generate access key.
Download GoOmni.mcpb from the same page and double-click it.
Paste your access key into the field Claude shows, then click Install. No config file, no Node install.

Option B — Manual config

Requires Node.js installed. The mcp-remote proxy bridges Claude's stdio transport to the hosted GoOmni endpoint.

Open ~/Library/Application Support/Claude/claude_desktop_config.json (create if it doesn't exist)
Add the GoOmni MCP server:

~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "goomni": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://agent.goomni.ai/api/mcp",
        "--transport", "http-only",
        "--header", "Authorization: Bearer <your-mcp-access-key>"
      ]
    }
  }
}

Fully quit and reopen Claude Desktop
Claude now has access to all GoOmni tools

Cursor IDE

AI-first code editor with MCP support

Cursor supports MCP servers natively. Add GoOmni and the AI agent inside Cursor can call any endpoint while you code: search docs, generate test cases, check compliance.

Setup

Open Cursor Settings → MCP → Add MCP Server
Add this configuration:

MCP Server config

{
  "mcpServers": {
    "goomni": {
      "url": "https://agent.goomni.ai/api/mcp",
      "headers": {
        "Authorization": "Bearer <your-mcp-access-key>"
      }
    }
  }
}

Cursor discovers all GoOmni tools
Ask the agent: "Search my GoOmni knowledge base for API docs"

ChatGPT Custom GPTs

OpenAI GPT Actions

Build a Custom GPT in ChatGPT that can call all GoOmni endpoints as 'actions.' ChatGPT reads the plugin manifest, discovers the OpenAPI spec, and every endpoint becomes a callable action.

Setup

Go to chat.openai.com → My GPTs → Create a GPT → Configure
Scroll to Actions → click Import from URL
Paste this URL:

Plugin manifest URL

https://agent.goomni.ai/.well-known/ai-plugin.json

ChatGPT reads the manifest, discovers the OpenAPI spec automatically
Every endpoint becomes a callable action in your Custom GPT
For authentication, set API Key with header name Cookie and paste your GoOmni session cookie (see the Authentication guide). GPT Actions call REST endpoints directly, which use the session cookie rather than a Bearer token.

Google Gemini

Gemini API function calling

Use GoOmni tools with Google's Gemini models via the Gemini API or Vertex AI. Fetch tool definitions in generic format and convert to Gemini's function declaration schema.

example_gemini.py

import requests
import google.generativeai as genai

genai.configure(api_key="YOUR_GEMINI_API_KEY")

# 1. Fetch GoOmni tools in Gemini-native format
data = requests.get(
    "https://agent.goomni.ai/api/docs/ai-tools?format=gemini"
).json()

# 2. Convert to Gemini function declarations
declarations = [
    genai.protos.FunctionDeclaration(**decl)
    for decl in data["function_declarations"]
]

# 3. Create model with GoOmni tools
model = genai.GenerativeModel(
    "gemini-3.7-flash",
    tools=[genai.protos.Tool(function_declarations=declarations)]
)

# 4. Gemini can now call GoOmni endpoints
chat = model.start_chat()
response = chat.send_message(
    "Generate QA test cases for project proj_abc123"
)
print(response.candidates[0].content.parts)

Native format: /api/docs/ai-tools?format=gemini. Also works with Vertex AI.

OpenAI Function Calling

For developers using the OpenAI API

Fetch pre-built tool definitions from GoOmni and pass them directly to openai.chat.completions.create(). The LLM decides which endpoint to call based on the user's request.

example.py

import requests
import openai

# 1. Fetch GoOmni tool definitions (all GoOmni tools)
tools = requests.get(
    "https://agent.goomni.ai/api/docs/ai-tools?format=openai"
).json()["tools"]

# 2. Pass tools to GPT, which can now call any GoOmni endpoint
response = openai.chat.completions.create(
    model="gpt-5.6",
    tools=tools,
    messages=[{
        "role": "user",
        "content": "Generate QA test cases for project proj_abc123"
    }]
)

# 3. GPT returns tool_calls → execute the HTTP request → return result
for tool_call in response.choices[0].message.tool_calls:
    print(f"Calling: {tool_call.function.name}")
    print(f"Args: {tool_call.function.arguments}")

Filter by domain: /api/docs/ai-tools?format=openai&tag=QA returns only QA-related tools.

Anthropic / Claude Function Calling

For developers using the Anthropic API

Same concept, Anthropic's tool_use format.

example.py

import requests
import anthropic

# Fetch in Anthropic format
tools = requests.get(
    "https://agent.goomni.ai/api/docs/ai-tools?format=anthropic"
).json()["tools"]

response = anthropic.messages.create(
    model="claude-sonnet-5",
    tools=tools,
    messages=[{
        "role": "user",
        "content": "Search the knowledge base for compliance requirements"
    }]
)

Any LLM / Custom Framework

LangChain, LlamaIndex, Gemini, or custom

Get generic tool definitions with HTTP method, path, and parameter schemas. Adapt to any framework.

Terminal

# Generic format
curl https://agent.goomni.ai/api/docs/ai-tools?format=generic

# Filter to specific domain
curl "https://agent.goomni.ai/api/docs/ai-tools?format=generic&tag=Voice"

Integration Endpoints

Endpoint	Purpose	Used By
POST /api/mcp	MCP JSON-RPC server (tools/list, tools/call)	Antigravity, Claude Desktop, Cursor
GET /api/mcp	MCP server info + setup instructions	Discovery
GET /.well-known/ai-plugin.json	ChatGPT Actions plugin manifest	ChatGPT Custom GPTs
GET /api/docs/ai-tools?format=openai	OpenAI function calling tool defs	OpenAI API
GET /api/docs/ai-tools?format=anthropic	Anthropic tool_use definitions	Anthropic API
GET /api/docs/ai-tools?format=gemini	Gemini function declarations	Gemini API, Vertex AI
GET /api/docs/ai-tools?format=generic	Generic tool definitions	LangChain, LlamaIndex, any LLM
GET /api/docs/openapi	Full OpenAPI 3.0 spec (JSON)	Swagger UI, code generators
GET /api/docs/postman	Postman collection download	Postman
/api-docs	Interactive API Reference	Developers