Drop-In Proxy - Curtly

Curtly provides a 100% OpenAI and Anthropic API-compatible reverse proxy. By simply pointing your client library’s baseURL (or base_url) to https://curtly.dev/v1, your requests are automatically passed through both optimization engines:

Input Stage: Prompts and context chunks are parsed in-memory via Safe Vault and compressed.
Output Stage: Ponytail brevity directives are injected to prevent model verbosity.
Forwarding: The optimized payload is forwarded to your target provider (OpenAI, Anthropic, OpenRouter, Groq, or custom endpoints) with sub-5ms latency overhead.

1-Line Integration Examples

from openai import OpenAI

# 1-Line change: update base_url
client = OpenAI(
    base_url="https://curtly.dev/v1",
    api_key="ctly_live_YOUR_CURTLY_KEY"
)

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": "You are a code reviewer."},
        {"role": "user", "content": "Review this pull request for race conditions."}
    ]
)
print(response.choices[0].message.content)

import OpenAI from 'openai';

// 1-Line change: update baseURL
const client = new OpenAI({
  baseURL: 'https://curtly.dev/v1',
  apiKey: process.env.CURTLY_API_KEY,
});

const response = await client.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Write an idempotent database migration script in SQL.' }],
});
console.log(response.choices[0].message.content);

import anthropic

# 1-Line change: update base_url
client = anthropic.Anthropic(
    base_url="https://curtly.dev",
    api_key="ctly_live_YOUR_CURTLY_KEY"
)

message = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Analyze this Kubernetes deployment manifest."}
    ]
)
print(message.content[0].text)

curl -X POST https://curtly.dev/v1/chat/completions \
  -H "Authorization: Bearer ctly_live_YOUR_CURTLY_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      { "role": "user", "content": "Explain how database indexing affects write throughput." }
    ]
  }'

Pass-Through Key Mode

You can also use Curtly in Pass-Through Mode by providing your raw provider key (sk-..., sk-ant-..., sk-or-v1-...). Curtly auto-detects the provider and forwards the request transparently using your own upstream quota without storing credentials.

curl -X POST https://curtly.dev/v1/chat/completions \
  -H "Authorization: Bearer sk-your-openai-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [{ "role": "user", "content": "Write a quicksort algorithm in Python." }]
  }'

Streaming (SSE) Support

Curtly fully supports server-sent events (stream: true) for both OpenAI and Anthropic protocols with zero buffering delay:

const stream = await client.chat.completions.create({
  model: 'gpt-4o-mini',
  messages: [{ role: 'user', content: 'Stream a 5-step deployment checklist.' }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content || '');
}

Telemetry Response Headers

Every proxy response includes transparent token savings headers:

HTTP/1.1 200 OK
Content-Type: application/json
X-Curtly-Original-Tokens: 384
X-Curtly-Input-Tokens: 198
X-Curtly-Input-Saved: 186
X-Curtly-Output-Saved-Est: 240
X-Curtly-Latency-Ms: 1.82
X-Curtly-Vault-Status: intact

⌘I