Custom client integration

Integrate any AI client or application with Antigravity Manager

Overview

Antigravity Manager supports any client that implements OpenAI, Anthropic, or Gemini API protocols. This guide covers how to integrate custom clients, applications, and scripts.

Supported protocols

Antigravity provides three API protocol formats:

OpenAI

Endpoint: /v1/chat/completions

Use for: Most third-party AI applications

Compatibility: 99% of AI tools

Anthropic

Endpoint: /v1/messages

Use for: Claude-native clients

Features: Full thinking mode support

Gemini

Endpoint: /v1/models/*:generateContent

Use for: Google AI SDK clients

Features: Native Gemini protocol

Basic configuration

All custom clients need three pieces of information:

1. Base URL

   Point your client to Antigravity’s proxy:

   http://127.0.0.1:8045
   

   For OpenAI-compatible clients, append /v1:

   http://127.0.0.1:8045/v1
   
2. API Key

   Use any value if authentication is disabled:

   sk-antigravity
   

   Or use your configured API key if authentication is enabled.

3. Model name

   Choose from available models:

   • gemini-3-flash - Fast, cost-effective
   • gemini-3-pro-high - High quality
   • claude-sonnet-4-6 - Balanced
   • claude-sonnet-4-6-thinking - Extended reasoning

OpenAI-compatible clients

Configuration template

Most OpenAI-compatible clients accept these settings:

{
  "baseURL": "http://127.0.0.1:8045/v1",
  "apiKey": "sk-antigravity",
  "defaultModel": "gemini-3-flash"
}

cURL examples

Basic chat

curl -X POST http://127.0.0.1:8045/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-antigravity" \
  -d '{
    "model": "gemini-3-flash",
    "messages": [
      {"role": "user", "content": "Hello!"}
    ]
  }'

Streaming

curl -X POST http://127.0.0.1:8045/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-antigravity" \
  -d '{
    "model": "gemini-3-flash",
    "messages": [
      {"role": "user", "content": "Write a story"}
    ],
    "stream": true
  }'

Image generation

curl -X POST http://127.0.0.1:8045/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-antigravity" \
  -d '{
    "model": "gemini-3-pro-image",
    "prompt": "A beautiful sunset",
    "size": "1024x1024",
    "quality": "hd"
  }'

Anthropic/Claude clients

Configuration

{
  "baseURL": "http://127.0.0.1:8045",
  "apiKey": "sk-antigravity"
}

cURL example

curl -X POST http://127.0.0.1:8045/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: sk-antigravity" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 1024,
    "messages": [
      {"role": "user", "content": "Hello!"}
    ]
  }'

Gemini clients

Configuration

{
  "baseURL": "http://127.0.0.1:8045",
  "apiKey": "sk-antigravity"
}

cURL example

curl -X POST "http://127.0.0.1:8045/v1/models/gemini-3-flash:generateContent" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-antigravity" \
  -d '{
    "contents": [{
      "parts": [{"text": "Hello!"}],
      "role": "user"
    }]
  }'

Popular applications

LangChain

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="http://127.0.0.1:8045/v1",
    api_key="sk-antigravity",
    model="gemini-3-flash"
)

response = llm.invoke("Hello!")
print(response.content)

LlamaIndex

from llama_index.llms.openai import OpenAI

llm = OpenAI(
    api_key="sk-antigravity",
    api_base="http://127.0.0.1:8045/v1",
    model="gemini-3-flash"
)

response = llm.complete("Hello!")
print(response)

Continue.dev

// config.json
{
  "models": [
    {
      "title": "Antigravity Gemini",
      "provider": "openai",
      "model": "gemini-3-flash",
      "apiBase": "http://127.0.0.1:8045/v1",
      "apiKey": "sk-antigravity"
    }
  ]
}

Cline (formerly Claude Dev)

{
  "apiProvider": "anthropic",
  "apiKey": "sk-antigravity",
  "apiUrl": "http://127.0.0.1:8045",
  "modelId": "claude-sonnet-4-6"
}

Authentication modes

Antigravity supports different authentication configurations:

Off (Development)

No authentication required:

curl http://127.0.0.1:8045/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{...}'

Only use this for local development. Not recommended for production.

Auto (Recommended)

Automatically enables auth for LAN access:

# Local (no auth needed)
curl http://127.0.0.1:8045/...

# Remote (auth required)
curl http://192.168.1.100:8045/... \
  -H "Authorization: Bearer sk-your-key"

Strict

Always requires authentication:

curl http://127.0.0.1:8045/v1/chat/completions \
  -H "Authorization: Bearer sk-your-key" \
  -d '{...}'

Advanced features

Model routing

Map client model names to Antigravity models:

// Antigravity model routing config
{
  "mappings": [
    {
      "from": "gpt-4",
      "to": "gemini-3-pro-high"
    },
    {
      "from": "gpt-3.5-turbo",
      "to": "gemini-3-flash"
    }
  ]
}

Now clients can use familiar names:

curl -X POST http://127.0.0.1:8045/v1/chat/completions \
  -H "Authorization: Bearer sk-antigravity" \
  -d '{
    "model": "gpt-4",
    "messages": [...]
  }'
# Automatically routes to gemini-3-pro-high

Session binding

For multi-turn conversations, enable session binding:

curl -X POST http://127.0.0.1:8045/v1/chat/completions \
  -H "Authorization: Bearer sk-antigravity" \
  -H "X-Session-ID: my-conversation-123" \
  -d '{...}'

Antigravity will use the same account for all requests with this session ID.

Custom headers

Pass custom headers for advanced control:

X-Preferred-Account · string

Use a specific account by email

X-Model-Tier · string

Prefer accounts of a specific tier (ultra, pro, free)

X-Session-ID · string

Enable sticky sessions for multi-turn conversations

Testing your integration

1. Test connectivity

   curl http://127.0.0.1:8045/health
   

   Expected: {"status": "ok"}

2. Test basic request

   curl -X POST http://127.0.0.1:8045/v1/chat/completions \
     -H "Content-Type: application/json" \
     -H "Authorization: Bearer sk-antigravity" \
     -d '{
       "model": "gemini-3-flash",
       "messages": [
         {"role": "user", "content": "Say hello"}
       ]
     }'
   
3. Test streaming

   curl -X POST http://127.0.0.1:8045/v1/chat/completions \
     -H "Content-Type: application/json" \
     -H "Authorization: Bearer sk-antigravity" \
     -d '{
       "model": "gemini-3-flash",
       "messages": [{"role": "user", "content": "Count to 10"}],
       "stream": true
     }'
   
4. Monitor in Antigravity

   Check the Monitor tab in Antigravity to see request logs and account rotation.

Troubleshooting

404 Not Found

Common causes:

• Wrong base URL (forgot /v1 for OpenAI protocol)
• Invalid model name
• Incorrect endpoint path

Solution: Verify endpoint matches protocol format

401 Unauthorized

Common causes:

• Authentication enabled but no key provided
• Wrong API key
• Missing Authorization header

Solution: Check auth mode in Antigravity settings

429 Rate Limited

Common causes:

• All accounts exhausted
• Quota protection enabled

Solution: Check dashboard, refresh quotas, add more accounts

Connection refused

Common causes:

• Antigravity proxy not running
• Wrong port
• Firewall blocking connection

Solution: Verify proxy is running and accessible

Best practices

1. Start simple: Test with cURL before integrating
2. Handle errors: Implement retry logic and error handling
3. Use streaming: For better UX with long responses
4. Monitor usage: Check Antigravity dashboard regularly
5. Configure routing: Map model names for easier migration
6. Enable session binding: For multi-turn conversations
7. Test thoroughly: Verify all features work as expected

Related documentation

• OpenAI API format
• Anthropic API format
• Gemini API format
• Authentication
• Model routing
• API Proxy settings