Anthropic Format - /v1/messages

Native Anthropic/Claude protocol endpoint with full thinking and tool support

Endpoint

POST http://127.0.0.1:8045/v1/messages

The native Anthropic/Claude protocol endpoint provides full support for Claude Code CLI and advanced features like extended thinking, system prompts, and tool use.

Authentication

x-api-key · string · required

Anthropic-style API key header

x-api-key: sk-antigravity

Alternatively, use standard Bearer authentication:

Authorization · string

Bearer token format

Authorization: Bearer sk-antigravity

Request Headers

Content-Type · string · required

Must be application/json

anthropic-version · string

API version, e.g., 2023-06-01

anthropic-beta · string

Beta features to enable. Automatically injected:

anthropic-beta: claude-code-20250219,interleaved-thinking-2025-05-14

Request Body

model · string · required

Model identifier:

• claude-sonnet-4-6 - Latest Claude Sonnet 4.6
• claude-sonnet-4-6-thinking - With extended thinking
• claude-opus-4-6-thinking - Opus with thinking
• gemini-3-pro-high - Mapped to Gemini (via protocol conversion)

messages · array · required

Conversation messages in Anthropic format

Show Message Object

role · string · required

Message role: user or assistant

content · string | array · required

Message content. Can be:

• String: Plain text
• Array: Multi-modal blocks (text, images, tool_use, tool_result)

system · string | array

System prompt to guide model behavior. Can be:

• String: Simple system message
• Array: Structured system blocks for advanced control

max_tokens · integer · required

Maximum tokens to generate. Required for Claude protocol.

Unlike OpenAI format, max_tokens is mandatory for Anthropic endpoints.

stream · boolean · default: "false"

Enable Server-Sent Events (SSE) streaming

temperature · number

Sampling temperature (0.0 to 1.0)

top_p · number

Nucleus sampling (0.0 to 1.0)

top_k · integer

Top-K sampling parameter

thinking · object

Extended thinking configuration for reasoning models

Show Thinking Config

type · string · required

Thinking mode:

• enabled - Enable with budget
• adaptive - Dynamic budget (maps to fixed budget internally)
• disabled - No extended thinking

budget_tokens · integer

Token budget for reasoning:

• Low: 8192
• Medium: 16384
• High: 24576

effort · string

Effort level: low, medium, high, max

output_config · object

Output configuration (Claude API v2.0.67+)

Show Output Config

effort · string

Model reasoning effort: low, medium, high

tools · array

Available tools for function calling

Show Tool Definition

name · string · required

Tool name

description · string

Tool description

input_schema · object · required

JSON Schema for tool parameters

Server tools like web_search can be enabled via:

Show Server Tool

type · string

Server tool type, e.g., web_search_20250305

name · string

Tool name: web_search

Response Format

Non-Streaming Response

id · string

Unique message identifier

type · string

Response type, always message

role · string

Always assistant

model · string

Model used for generation

content · array

Response content blocks

Show Content Blocks

type · string

Block type: text, thinking, tool_use, image

text · string

Text content (for text blocks)

thinking · string

Reasoning content (for thinking blocks)

signature · string

Cryptographic signature for thinking/tool_use blocks

id · string

Tool call ID (for tool_use blocks)

name · string

Tool name (for tool_use blocks)

input · object

Tool parameters (for tool_use blocks)

stop_reason · string

Completion reason: end_turn, max_tokens, tool_use, stop_sequence

usage · object

Token usage statistics

Show Usage Object

input_tokens · integer

Prompt tokens

output_tokens · integer

Completion tokens

cache_read_input_tokens · integer

Cached tokens read (prompt caching)

cache_creation_input_tokens · integer

Tokens cached for future use

Example: Basic Chat

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

Example: With System Prompt

curl -X POST http://127.0.0.1:8045/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: sk-antigravity" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 2048,
    "system": "You are a helpful coding assistant. Always explain your reasoning.",
    "messages": [
      {"role": "user", "content": "Write a Python function to reverse a string"}
    ]
  }'

Example: Extended Thinking

curl -X POST http://127.0.0.1:8045/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: sk-antigravity" \
  -d '{
    "model": "claude-sonnet-4-6-thinking",
    "max_tokens": 4096,
    "thinking": {
      "type": "enabled",
      "budget_tokens": 16384
    },
    "messages": [
      {"role": "user", "content": "Solve this complex logic puzzle: ..."}
    ]
  }'

Example: Claude Code CLI Integration

# Set environment variables
export ANTHROPIC_API_KEY="sk-antigravity"
export ANTHROPIC_BASE_URL="http://127.0.0.1:8045"

# Run Claude Code CLI
claude "Help me debug this Python code"

Example: Tool Use

curl -X POST http://127.0.0.1:8045/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: sk-antigravity" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 2048,
    "tools": [
      {
        "name": "get_weather",
        "description": "Get current weather",
        "input_schema": {
          "type": "object",
          "properties": {
            "location": {"type": "string"}
          },
          "required": ["location"]
        }
      }
    ],
    "messages": [
      {"role": "user", "content": "What's the weather in Tokyo?"}
    ]
  }'

Example: Web Search (Server Tool)

curl -X POST http://127.0.0.1:8045/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: sk-antigravity" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 2048,
    "tools": [
      {
        "type": "web_search_20250305",
        "name": "web_search"
      }
    ],
    "messages": [
      {"role": "user", "content": "What are the latest AI developments?"}
    ]
  }'

Content Block Types

Text Block

{
  "type": "text",
  "text": "Here is my response..."
}

Thinking Block

{
  "type": "thinking",
  "thinking": "Let me analyze this step by step...",
  "signature": "base64-encoded-signature"
}

Tool Use Block

{
  "type": "tool_use",
  "id": "toolu_123abc",
  "name": "get_weather",
  "input": {
    "location": "Tokyo"
  },
  "signature": "base64-encoded-signature"
}

Tool Result Block (in messages)

{
  "role": "user",
  "content": [
    {
      "type": "tool_result",
      "tool_use_id": "toolu_123abc",
      "content": "Temperature: 22°C, Sunny"
    }
  ]
}

Image Upload

{
  "role": "user",
  "content": [
    {
      "type": "image",
      "source": {
        "type": "base64",
        "media_type": "image/jpeg",
        "data": "base64-encoded-image-data"
      }
    },
    {
      "type": "text",
      "text": "What's in this image?"
    }
  ]
}

Features

• Full Claude Protocol: 100% compatible with Claude Code CLI and Anthropic SDKs
• Extended Thinking: Support for reasoning tokens with signature validation
• Tool Use: Client tools, server tools (web_search), and MCP integration
• Streaming: Real-time SSE streaming with thinking blocks
• Context Caching: Automatic prompt caching for repeated contexts
• Multi-modal: Images, documents, audio support

Error Format

{
  "type": "error",
  "error": {
    "type": "invalid_request_error",
    "message": "max_tokens is required"
  }
}

Common error types:

• invalid_request_error - Malformed request
• authentication_error - Invalid API key
• rate_limit_error - Quota exceeded
• server_error - Internal error