Prompt caching
Pattern
BEG·Jun 26

Anthropic Cache Control (AI Gateway)

Lightweight ToolLoopAgent pattern for Anthropic ephemeral cache control through Vercel AI Gateway: prepareCall injects a cached system breakpoint and reports cache read/write tokens.

View Full Pattern on Desktop
Anthropic Cache Control (AI Gateway) preview

Problems Solved

  • Anthropic cacheControl breakpoints via AI Gateway
  • Inject cached system prefix with prepareCall on ToolLoopAgent
  • Measure cache creation vs cache read tokens per turn

Use Cases

  • Explicit ephemeral caching for long agent system prompts
  • Multi-turn agents with stable policy prefixes
  • Teach cache read vs cache write metrics

AI SDK APIs

ToolLoopAgentgatewayproviderOptionsprepareCallstepCountIs

Tags

aiagentsprompt-cacheanthropiccache-controlai-gatewayprepareCall

Requirements

Node.js
24+
Memory
256MB
Other
  • Vercel AI Gateway API key

Environment Variables

External Services

anthropic

Ready to use?

View the full interactive pattern with live preview and copy the source code.

View Pattern on Desktop

Related Patterns

OpenAI Prompt Caching (AI Gateway) preview

OpenAI Prompt Caching (AI Gateway)

Lightweight ToolLoopAgent pattern for OpenAI prompt caching through Vercel AI Gateway: stable system instructions, promptCacheKey, and per-turn cachedPromptTokens metrics.

aiagents+5
View →
Sub-Agent Orchestrator preview

Sub-Agent Orchestrator

Custom Agent implementation demonstrating the Agent interface abstraction with an orchestrator that routes queries to specialized sub-agents (research, analysis, support). Shows options passing and structured outputs.

aiagents+7
View →
Just Bash Skills Agent preview

Just Bash Skills Agent

Minimal ToolLoopAgent chat demo: on-call-style skills (incident triage + stakeholder comms) on disk, skill tool loads SKILL.md + paths, readFile pulls reference checklists and templates from the sandbox.

aiagents+5
View →
Call Options: prepareCall preview

Call Options: prepareCall

Dynamically configure one ToolLoopAgent per request with callOptionsSchema and prepareCall. A customer-support agent rewrites instructions, gates tools, tunes reasoning effort, and injects RAG context based on account tier and urgency.

aiagents+7
View →
End of Pattern