Breaking

Autocache proxy auto-injects cache headers into Claude API calls

July 2, 2026 at 18:29 EDT

AI Agents
Open Source
Infra & Chips

montevive has released Autocache, an open-source proxy that automatically injects cache_control headers into Anthropic Claude API requests. It enables Prompt Caching without code changes and slots transparently into existing tools such as n8n and LangChain.

Open-Source Release · Autocache

A Transparent Proxy That Cuts Claude API Costs by up to 90% — With Zero Code Changes

Autocache auto-injects cache_control headers into Anthropic Claude requests, unlocking Prompt Caching for no-code tools like n8n and Flowise. Just repoint your API base URL — a true drop-in replacement.

90%

Max cost reduction per request

85%

Max latency reduction

~29MB

Docker image (amd64/arm64)

COST PER REQUEST · 15,000-token agent

First call vs. every cached call after it

$0.045

First call
(uncached)

→

$0.0045

Cached calls
(10× cheaper)

Repeated agent calls reuse cached system prompts & tool definitions.

Cache Strategies — set via CACHE_STRATEGY

conservative

max breakpoints · System / tools focus

moderate (default)

max breakpoints · Standard setup

aggressive

max breakpoints · Maximum caching

Why it lands well

Works as-is with n8n, Flowise, Make.com, LangChain, LlamaIndex
No manual cache_control editing
Real-time ROI analytics in response headers
Self-hosted, multi-tenant, streaming supported

The caveats

Benefit fades with lots of dynamic content
Below-minimum tokens may not cache at all
Claude on Bedrock / Google Cloud may lack auto-caching
Just released — real-world evaluation still limited

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.