Repository Public
1---2title: Ruby SDK3url: https://platform.claude.com/docs/en/cli-sdks-libraries/sdks/ruby4description: Install and configure the Anthropic Ruby SDK with Sorbet types, streaming helpers, and connection pooling5---67The Anthropic Ruby library provides convenient access to the Claude API from any Ruby 3.2.0+ application. It ships with comprehensive types and docstrings in Yard, RBS, and RBI. The standard library's `net/http` is used as the HTTP transport, with connection pooling through the `connection_pool` gem.89<Info>10 For API feature documentation with code examples, see the [API reference](https://platform.claude.com/docs/en/api/overview). This page covers Ruby-specific SDK features and configuration.11</Info>1213## Installation1415Add the gem to your application's `Gemfile` with Bundler:1617```bash18bundle add anthropic19```2021## Requirements2223Ruby 3.2.0 or higher.2425## Usage2627```ruby28anthropic = Anthropic::Client.new(29 api_key: ENV["ANTHROPIC_API_KEY"] # This is the default and can be omitted30)3132message = anthropic.messages.create(33 max_tokens: 1024,34 messages: [{role: "user", content: "Hello, Claude"}],35 model: :"claude-opus-5"36)3738message.content.each do |block|39 puts block.text if block.type == :text40end41```4243For authentication options including Workload Identity Federation, see [Authentication](https://platform.claude.com/docs/en/manage-claude/authentication). If your API key is a [personal or service account key](https://platform.claude.com/docs/en/manage-claude/authentication#key-types) with access to multiple workspaces, set the workspace ID in the `anthropic-workspace-id` request header; [Select a workspace](https://platform.claude.com/docs/en/manage-claude/authentication#select-a-workspace) shows the per-request option for this SDK.4445## Streaming4647The SDK provides support for streaming responses using Server-Sent Events (SSE).4849```ruby50anthropic = Anthropic::Client.new51stream = anthropic.messages.stream(52 max_tokens: 1024,53 messages: [{role: "user", content: "Hello, Claude"}],54 model: :"claude-opus-5"55)5657stream.each do |message|58 puts(message.type)59end60```6162### Streaming helpers6364This library provides several conveniences for streaming messages, for example:6566```ruby67anthropic = Anthropic::Client.new68stream = anthropic.messages.stream(69 max_tokens: 1024,70 messages: [{role: :user, content: "Say hello there!"}],71 model: :"claude-opus-5"72)7374stream.text.each do |text|75 print(text)76end77```7879Streaming with `anthropic.messages.stream(...)` exposes various helpers including accumulation and SDK-specific events.8081## Input schema and tool calling8283The SDK provides helper mechanisms to define structured data classes for tools and let Claude automatically execute them. For detailed documentation on tool use patterns including the tool runner, see [Tool Runner (SDK)](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-runner).8485```ruby86anthropic = Anthropic::Client.new87class CalculatorInput < Anthropic::BaseModel88 required :lhs, Float89 required :rhs, Float90 required :operator, Anthropic::InputSchema::EnumOf[:+, :-, :*, :/]91end9293class Calculator < Anthropic::BaseTool94 input_schema CalculatorInput9596 def call(expr)97 expr.lhs.public_send(expr.operator, expr.rhs)98 end99end100101# Automatically handles tool execution loop102anthropic.beta.messages.tool_runner(103 model: "claude-opus-5",104 max_tokens: 1024,105 messages: [{role: "user", content: "What's 15 * 7?"}],106 tools: [Calculator.new]107).each_message { |message| puts message.content }108```109110## Structured outputs111112For complete structured outputs documentation including Ruby examples, see [Structured outputs](https://platform.claude.com/docs/en/build-with-claude/structured-outputs).113114## Handling errors115116When the library is unable to connect to the API, or if the API returns a non-success status code (that is, 4xx or 5xx response), a subclass of `Anthropic::Errors::APIError` is raised:117118```ruby119anthropic = Anthropic::Client.new120begin121 message = anthropic.messages.create(122 max_tokens: 1024,123 messages: [{role: "user", content: "Hello, Claude"}],124 model: :"claude-opus-5"125 )126rescue Anthropic::Errors::APIConnectionError => e127 puts("The server could not be reached")128 puts(e.cause) # an underlying Exception, likely raised within `net/http`129rescue Anthropic::Errors::RateLimitError => e130 puts("A 429 status code was received; we should back off a bit.")131rescue Anthropic::Errors::APIStatusError => e132 puts("Another non-200-range status code was received")133 puts(e.status)134end135```136137Error codes are as follows:138139| Cause | Error Type |140| ---------------- | -------------------------- |141| HTTP 400 | `BadRequestError` |142| HTTP 401 | `AuthenticationError` |143| HTTP 403 | `PermissionDeniedError` |144| HTTP 404 | `NotFoundError` |145| HTTP 409 | `ConflictError` |146| HTTP 422 | `UnprocessableEntityError` |147| HTTP 429 | `RateLimitError` |148| HTTP >= 500 | `InternalServerError` |149| Other HTTP error | `APIStatusError` |150| Timeout | `APITimeoutError` |151| Network error | `APIConnectionError` |152153## Retries154155Certain errors will be automatically retried 2 times by default, with a short exponential backoff.156157Connection errors (for example, because of a network connectivity problem), 408 Request Timeout, 409 Conflict, 429 Rate Limit, >=500 Internal errors, and timeouts are all retried by default.158159You can use the `max_retries` option to configure or disable this:160161```ruby162# Configure the default for all requests:163anthropic = Anthropic::Client.new(164 max_retries: 0 # default is 2165)166167# Or, configure per-request:168anthropic.messages.create(169 max_tokens: 1024,170 messages: [{role: "user", content: "Hello, Claude"}],171 model: :"claude-opus-5",172 request_options: {max_retries: 5}173)174```175176## Timeouts177178By default, requests time out after 10 minutes. You can use the `timeout` option to configure this:179180```ruby181# Configure the default for all requests:182anthropic = Anthropic::Client.new(183 timeout: 20 # 20 seconds (default is 10 minutes)184)185186# Or, configure per-request:187anthropic.messages.create(188 max_tokens: 1024,189 messages: [{role: "user", content: "Hello, Claude"}],190 model: :"claude-opus-5",191 request_options: {timeout: 5}192)193```194195On timeout, `Anthropic::Errors::APITimeoutError` is raised.196197Note that requests that time out are retried by default.198199## Pagination200201List methods in the Claude API are paginated.202203This library provides auto-paginating iterators with each list response, so you do not have to request successive pages manually:204205```ruby206anthropic = Anthropic::Client.new207page = anthropic.messages.batches.list(limit: 20)208209# Fetch single item from page.210batch = page.data[0]211puts(batch.id)212213# Automatically fetches more pages as needed.214page.auto_paging_each do |batch|215 puts(batch.id)216end217```218219Alternatively, you can use the `#next_page?` and `#next_page` methods for more granular control working with pages.220221```ruby222anthropic = Anthropic::Client.new223page = anthropic.messages.batches.list(limit: 20)224loop do225 page.data&.each { |batch| puts(batch.id) }226 break unless page.next_page?227 page = page.next_page228end229```230231## File uploads232233Request parameters that correspond to file uploads can be passed as raw contents, a [`Pathname`](https://rubyapi.org/3.2/o/pathname) instance, [`StringIO`](https://rubyapi.org/3.2/o/stringio), or more.234235```ruby236anthropic = Anthropic::Client.new237require "pathname"238239# Use `Pathname` to send the filename and/or avoid paging a large file into memory:240file_metadata = anthropic.files.upload(file: Pathname("/path/to/file"))241242# Alternatively, pass file contents or a `StringIO` directly:243file_metadata = anthropic.files.upload(file: File.read("/path/to/file"))244245# Or, to control the filename and/or content type:246file = Anthropic::FilePart.new(File.read("/path/to/file"), filename: "/path/to/file", content_type: "...")247file_metadata = anthropic.files.upload(file: file)248249puts(file_metadata.id)250```251252Note that you can also pass a raw `IO` descriptor, but this disables retries, as the library can't be sure if the descriptor is a file or pipe (which cannot be rewound).253254## Sorbet255256This library provides comprehensive [RBI](https://sorbet.org/docs/rbi) definitions, and has no dependency on sorbet-runtime.257258You can provide typesafe request parameters like so:259260```ruby261anthropic = Anthropic::Client.new262anthropic.messages.create(263 max_tokens: 1024,264 messages: [Anthropic::MessageParam.new(role: "user", content: "Hello, Claude")],265 model: :"claude-opus-5"266)267```268269Or, equivalently:270271```ruby272anthropic = Anthropic::Client.new273# Hashes work, but are not typesafe:274anthropic.messages.create(275 max_tokens: 1024,276 messages: [{role: "user", content: "Hello, Claude"}],277 model: :"claude-opus-5"278)279280# You can also splat a full Params class:281params = Anthropic::MessageCreateParams.new(282 max_tokens: 1024,283 messages: [Anthropic::MessageParam.new(role: "user", content: "Hello, Claude")],284 model: :"claude-opus-5"285)286anthropic.messages.create(**params)287```288289### Enums290291Since this library does not depend on `sorbet-runtime`, it cannot provide [`T::Enum`](https://sorbet.org/docs/tenum) instances. Instead, the SDK provides "tagged symbols", which is always a primitive at runtime:292293```ruby294# :auto295puts(Anthropic::MessageCreateParams::ServiceTier::AUTO)296297# Revealed type: `T.all(Anthropic::MessageCreateParams::ServiceTier, Symbol)`298T.reveal_type(Anthropic::MessageCreateParams::ServiceTier::AUTO)299```300301Enum parameters have a "relaxed" type, so you can either pass in enum constants or their literal value:302303```ruby304# Using the enum constants preserves the tagged type information:305anthropic.messages.create(306 service_tier: Anthropic::MessageCreateParams::ServiceTier::AUTO,307 # ...308)309310# Literal values are also permissible:311anthropic.messages.create(312 service_tier: :auto,313 # ...314)315```316317## BaseModel318319All parameter and response objects inherit from `Anthropic::Internal::Type::BaseModel`, which provides several conveniences, including:3203211. All fields, including unknown ones, are accessible with `obj[:prop]` syntax, and can be destructured with `obj => {prop: prop}` or pattern-matching syntax.3223232. Structural equivalence for equality; if two API calls return the same values, comparing the responses with == will return true.3243253. Both instances and the classes themselves can be pretty-printed.3263274. Helpers such as `#to_h`, `#deep_to_h`, `#to_json`, and `#to_yaml`.328329## Concurrency and connection pooling330331The `Anthropic::Client` instances are threadsafe, but are only fork-safe when there are no in-flight HTTP requests.332333Each instance of `Anthropic::Client` has its own HTTP connection pool with a default size of 99. As such, the recommendation is to create the client once per application in most settings.334335When all available connections from the pool are checked out, requests wait for a new connection to become available, with queue time counting toward the request timeout.336337Unless otherwise specified, other classes in the SDK do not have locks protecting their underlying data structure.338339## Making custom or undocumented requests340341### Undocumented properties342343You can send undocumented parameters to any endpoint, and read undocumented response properties, like so:344345<Warning>346 The `extra_` parameters of the same name override the documented parameters. For security reasons, ensure these methods are only used with trusted input data.347</Warning>348349```ruby350anthropic = Anthropic::Client.new351value = "example"352message =353 anthropic.messages.create(354 max_tokens: 1024,355 messages: [{role: "user", content: "Hello, Claude"}],356 model: :"claude-opus-5",357 request_options: {358 extra_query: {my_query_parameter: value},359 extra_body: {my_body_parameter: value},360 extra_headers: {"my-header": value}361 }362 )363364puts(message[:my_undocumented_property])365```366367### Undocumented request params368369If you want to explicitly send an extra param, you can do so with the `extra_query`, `extra_body`, and `extra_headers` under the `request_options:` parameter when making a request, as seen in the examples above.370371### Undocumented endpoints372373To make requests to undocumented endpoints while retaining the benefit of auth, retries, and so on, you can make requests using `anthropic.request`, like so:374375```ruby376response = anthropic.request(377 method: :post,378 path: '/undocumented/endpoint',379 query: {"dog": "woof"},380 headers: {"useful-header": "interesting-value"},381 body: {"hello": "world"}382)383```384385## Platform integrations386387<Note>388 For detailed platform setup guides with code examples, see:389390 * [Amazon Bedrock](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock)391 * [Amazon Bedrock (Opus 4.6 and earlier)](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy)392 * [Claude Platform on AWS](https://platform.claude.com/docs/en/build-with-claude/claude-platform-on-aws)393 * [Google Cloud](https://platform.claude.com/docs/en/build-with-claude/claude-on-vertex-ai)394</Note>395396The Ruby SDK supports the following platforms:397398* **Agent Platform:** `Anthropic::VertexClient`. Requires the `googleauth` gem.399* **Bedrock:** `Anthropic::BedrockMantleClient`, or `Anthropic::BedrockClient` for the `bedrock-runtime` path. `Anthropic::BedrockMantleClient` requires the `aws-sdk-core` gem; `Anthropic::BedrockClient` requires the `aws-sdk-bedrockruntime` gem.400* **Claude Platform on AWS:** Part of the main `anthropic` gem (requires the `aws-sdk-core` gem). Provides `Anthropic::AWSClient`. Pass `workspace_id:` to the constructor or set the `ANTHROPIC_AWS_WORKSPACE_ID` environment variable (see [Workspaces](https://platform.claude.com/docs/en/build-with-claude/claude-platform-on-aws#workspaces)). Available in beta.401* **Foundry:** Not currently supported in the Ruby SDK. See [Claude in Microsoft Foundry](https://platform.claude.com/docs/en/build-with-claude/claude-in-microsoft-foundry) for supported SDKs.402403Use `Anthropic::BedrockMantleClient` for new projects; `Anthropic::BedrockClient` remains for existing applications using the Bedrock `InvokeModel` API.404405## Semantic versioning406407This package follows [SemVer](https://semver.org/spec/v2.0.0.html) conventions.408409This package considers improvements to the (non-runtime) `*.rbi` and `*.rbs` type definitions to be non-breaking changes.410411## Additional resources412413* [GitHub repository](https://github.com/anthropics/anthropic-sdk-ruby)414* [YARD documentation](https://gemdocs.org/gems/anthropic)415* [API reference](https://platform.claude.com/docs/en/api/overview)416* [Streaming Messages](https://platform.claude.com/docs/en/build-with-claude/streaming)417