Loading pattern preview
BlocksAgent Patterns
Anthropic Cache Control (AI Gateway)
Anthropic Cache Control (AI Gateway)
Lightweight ToolLoopAgent pattern for Anthropic ephemeral cache control through Vercel AI Gateway: cached system instructions with cacheControl and reports cache read/write tokens.
Use this when you need to set Anthropic cache breakpoints on a ToolLoopAgent's instructions and compare cache write versus cache read tokens per turn.
Loading pattern preview
Problems solved
Technical challenges this implementation handles out of the box.
- Set Anthropic ephemeral cacheControl breakpoints via AI Gateway
- Cache long system instructions on a ToolLoopAgent
- Measure cache creation versus cache read tokens per turn
Use cases
Products and workflows this pattern is designed to support.
- Explicit ephemeral caching for long agent policies
- Multi-turn agents with stable system prefixes
- Teaching Anthropic cache read versus write metrics
Compare with
Similar patterns, and when each one is the better fit.
OpenAI Prompt Caching (AI Gateway)Use this when your OpenAI agent resends the same system prompt each turn and you want to cache it through AI Gateway using promptCacheKey.
Agent Routing PatternUse this when one chat endpoint should sort queries into specialist agents and fall back across OpenAI, Perplexity, and Deepseek if a provider fails.
Jev MemoryUse this when an agent's memory store fills with stale or trivial facts and you need gated writes, budgeted recall, and eviction instead of rewriting.
Eve Long-Term Memory AgentUse this when your Eve agent needs per-user memory it can save, list, and forget, injected as context before every turn and isolated by tenant.
Setup
Requirements, wiring steps, and what this pattern adds to your project.
Getting started
Pick how you want to pull this pattern in. Then wire env vars and routes the same way.
- 1
Download → Next.js app
Use Download → Next.js app in the preview toolbar. You get a zip scaffold (Pro downloads rewrite preview `/view/.../api/*` paths to `/api/*` automatically).
- 2
Extract the zip and install dependencies
Unzip the archive, then run `pnpm install` (or npm / yarn) in the project folder.
- 3
Fill .env.local
The download includes .env.local placeholders when needed. Add real values for: AI_GATEWAY_API_KEY.
- 4
Customize the agent and tool files
Adapt prompts, tools, and stop conditions in the generated project for your product.
- 5
Run the app and open the pattern route
Run `pnpm dev`, open the setup/checklist page if present, then navigate to the generated pattern route and verify responses.
Environment variables
AI_GATEWAY_API_KEYGet key
Capabilities
AI SDK APIs
ToolLoopAgentgatewayproviderOptionsinstructionsstepCountIsProviders
Claude
Files added
9 files
app/page.tsxapp/layout.tsxapp/api/cache-anthropic/route.tslib/cache-anthropic-agent.tslib/build-cached-instructions.tslib/cache-metrics.tslib/stable-system-prompt.tslib/request-utils.tsREADME.md
Dependencies
9 total
npm packages
5AISDKlucide-reactNext.jsReactZod
Registry components
alertbadgebuttonskeleton
Critical files
2 files
lib/build-cached-instructions.tsBuilds SystemModelMessage instructions with providerOptions.anthropic.cacheControl ephemeral breakpoint.
lib/cache-anthropic-agent.tsToolLoopAgent uses buildAnthropicCachedInstructions() as instructions (not messages).
Requirements
- Complexity
- Intermediate
- Setup time
- ~5 minutes
- Node
- 24+
- Memory
- 256MB
