logo
Sign In

Anthropic Claude API Connection

function

A Secure Anthropic Claude API Pipeline

1337hero
@1337hero
·
a year ago
270
3
3.4K
Updated a year ago
Created Nov 13, 2025
anthropic_claude_api_connection.py
Python · 454 lines
"""
title: Anthropic Claude API
author: 1337Hero (Mike Key)
version: 2.1.0
license: MIT
required_open_webui_version: 0.5.0
requirements: requests>=2.31.0

A clean, simple Anthropic Claude integration that does exactly what's needed.
Security by default, not security theater.

Features:

Anthropic Claude API Pipeline

What's New in v2.1

🚀 Dynamic Model List - Now automatically fetches available models from Anthropic's API
🔄 Auto-Refresh - Model list updates hourly (configurable)
🛡️ Graceful Fallback - Uses cached/fallback models if API is unavailable

No more waiting for pipeline updates when Anthropic releases new models!

Overview

A clean, secure Anthropic Claude integration for Open WebUI. Built with DHH-inspired principles: simple, expressive, and secure by default—not security
theater.

This pipeline provides seamless access to all Claude models with essential security features and zero configuration overhead.

Supported Models

Models are automatically fetched from Anthropic's /v1/models API. New models are picked up automatically when Anthropic releases them.

Fallback Models (used if API unavailable)

  • claude-sonnet-4-5 - Best balance of intelligence, speed, and cost
  • claude-haiku-4-5 - Fastest responses with near-frontier intelligence
  • claude-opus-4-1 - Specialized reasoning for complex tasks
  • claude-3-7-sonnet - Enhanced capabilities with extended thinking
  • claude-3-5-sonnet - Strong general-purpose model
  • claude-3-5-haiku - Fast and cost-effective
  • claude-3-opus - Previous generation flagship
  • claude-3-sonnet - Balanced performance
  • claude-3-haiku - Speed-optimized

Features

Security (Real Protection, Not Theater)

✅ SSRF Protection - Validates URLs, blocks private networks and metadata endpoints
✅ Input Validation - Prevents crashes from malformed data
✅ Secure Error Handling - User-friendly messages, no information disclosure
✅ Image Size Validation - Enforces Anthropic's 5MB limit per image

Core Functionality

✅ Dynamic Model List - Auto-fetches available models from Anthropic's API
✅ Auto-Refresh - Configurable refresh interval (default: 1 hour)
✅ Streaming Support - Real-time response streaming
✅ Multimodal - Text and image processing
✅ Retry Logic - Automatic retry with exponential backoff
✅ Graceful Fallback - Uses cached/fallback models if API is unavailable

Configuration

Simple Configuration (Two Valves)

  • ANTHROPIC_API_KEY - Your Anthropic API key (required)
  • MODEL_REFRESH_INTERVAL - Seconds between model list refreshes, 0 to disable (default: 3600)

Minimal config, maximum functionality.

License

MIT License

Contributing

Github: https://github.com/1337hero/open-webui-anthropic-api-pipe

Contributions welcome! But remember:

Before adding features, ask:

  1. Is this solving an actual problem or a theoretical one?
  2. Can we solve this with less code?
  3. Are we duplicating Anthropic's existing protections?

Pull Request Guidelines:

  • Keep it simple
  • Remove code when possible
  • No security theater
  • Test with real use cases

Key changes:

  • Added "What's New in v2.1" section at top
  • Updated "Supported Models" to explain dynamic fetching
  • Added 3 new features to Core Functionality
  • Added MODEL_REFRESH_INTERVAL to Configuration
0