anthropic.twins.la

A digital twin of the Anthropic Messages API.

What is this?

A high-fidelity digital twin of the public Anthropic Messages API. The twin returns synthetic, deterministic responses that match the real API's shape exactly — consumer SDKs (anthropic, etc.) work without modification.

Use it to prototype, run CI, or test integrations without burning live credits or hitting quotas.

Supported scenarios

messages-api — POST /v1/messages (full + SSE streaming), POST /v1/messages/count_tokens, GET /v1/models[/{id}], plus prompt-caching-2024-07-31 and extended-thinking beta headers.

How to use it

Cloud: Mint an api key via POST /_twin/accounts on https://anthropic.twins.la, then point your Anthropic SDK at the twin's base URL.

Local: Install with pip install twins-anthropic-local and run a local instance on any port. Same API, same behaviour, your machine.

For agents

Copy this into your agent's system prompt, tool configuration, or CLAUDE.md. Also available as plain text at /_twin/agent-instructions.

# Anthropic Messages API Twin — anthropic.twins.la

A digital twin of the public Anthropic Messages API. Synthetic responses
only — the twin never calls api.anthropic.com. The shape matches the
real API verbatim, so consumer SDKs work without modification.

## Authentication

Twin Plane: HTTP Basic (tenant_id:tenant_secret)
  Bootstrap a tenant first:
    POST /_twin/tenants -> {tenant_id, tenant_secret}

Twin Plane Admin: Bearer token (set by deployment owner)
  Authorization: Bearer <admin_token>
  Or: X-Twin-Admin-Token: <admin_token>

Provider (data plane):
  x-api-key: sk-ant-twin-...    (mint via POST /_twin/accounts)
  anthropic-version: 2023-06-01 (any value accepted)

  Optional:
  anthropic-beta: prompt-caching-2024-07-31
  anthropic-beta: extended-thinking-2025-01-15

## Key Endpoints

Twin Plane (no auth):
  GET  /_twin/health
  GET  /_twin/scenarios
  GET  /_twin/settings
  GET  /_twin/references
  GET  /_twin/agent-instructions
  POST /_twin/tenants

Twin Plane (Basic tenant_id:tenant_secret):
  POST /_twin/accounts            — kind=api_key — returns {api_key} once
  GET  /_twin/accounts            — list api keys (masked)
  GET  /_twin/messages            — message history for this tenant
  GET  /_twin/logs
  POST /_twin/feedback
  GET  /_twin/feedback

Provider (x-api-key + anthropic-version):
  POST /v1/messages               — full or SSE streaming via stream:true
  POST /v1/messages/count_tokens
  GET  /v1/models
  GET  /v1/models/<model_id>

## Quick Start (local)

1. pip install twins-anthropic twins-anthropic-local
   python -m twins_anthropic_local

2. Bootstrap a tenant:
   curl -X POST http://localhost:8080/_twin/tenants \
     -H "Content-Type: application/json" \
     -d '{"friendly_name": "Dev"}'
   # -> { tenant_id, tenant_secret }

3. Mint an api key:
   curl -X POST http://localhost:8080/_twin/accounts \
     -u "TENANT_ID:TENANT_SECRET" \
     -H "Content-Type: application/json" \
     -d '{"kind":"api_key","friendly_name":"Demo Key"}'
   # -> { api_key, api_key_id, ... }

4. Call the Messages API:
   curl -X POST http://localhost:8080/v1/messages \
     -H "x-api-key: $API_KEY" \
     -H "anthropic-version: 2023-06-01" \
     -H "Content-Type: application/json" \
     -d '{"model":"claude-3-5-sonnet-latest","max_tokens":64,
          "messages":[{"role":"user","content":"hi"}]}'
   # -> { id, type:"message", role:"assistant", content:[{type:"text",...}], ... }

## Reference

GitHub:           https://github.com/twins-la/anthropic
Project overview: https://twins.la
All twins:        https://github.com/twins-la/twins-la

GitHub twins.la Health Scenarios