Anthropic

Auto-detected#

Delimiter automatically detects requests to api.anthropic.com and reads rate-limit headers from every response. No setup needed beyond delimiter.init().

Supported headers#

Anthropic returns rate limit information on every API response:

anthropic-ratelimit-requests-limit: 4000
anthropic-ratelimit-requests-remaining: 3201
anthropic-ratelimit-tokens-limit: 400000
anthropic-ratelimit-tokens-remaining: 312000
anthropic-ratelimit-requests-reset: 2025-01-15T14:24:00Z
anthropic-ratelimit-tokens-reset: 2025-01-15T14:23:15Z

What gets parsed#

| Header | Parsed Field | | --- | --- | | anthropic-ratelimit-requests-limit | requests_limit | | anthropic-ratelimit-requests-remaining | requests_remaining | | anthropic-ratelimit-tokens-limit | tokens_limit | | anthropic-ratelimit-tokens-remaining | tokens_remaining | | anthropic-ratelimit-requests-reset | reset_requests_ms (parsed from ISO 8601) | | anthropic-ratelimit-tokens-reset | reset_tokens_ms (parsed from ISO 8601) |

Usage#

import { delimiter } from '@delimiter/sdk'
import Anthropic from '@anthropic-ai/sdk'

delimiter.init('dlm_key')

// Use Anthropic as normal — Delimiter monitors automatically
const anthropic = new Anthropic({ apiKey: process.env.ANTHROPIC_KEY })

await anthropic.messages.create({
  model: 'claude-sonnet-4-20250514',
  max_tokens: 1024,
  messages: [{ role: 'user', content: 'Hello' }]
})

Works the same with LangChain, Vercel AI SDK, or raw fetch('https://api.anthropic.com/...') calls.