Class AnthropicProvider

AnthropicProvider: Concrete BaseProvider implementation for the Anthropic API.

Responsibilities:

Owns the Anthropic SDK client
Initializes provider-level configuration
Registers supported capabilities
Delegates capability execution to specialized implementations

Implements chat, embedding, and moderation capabilities for Anthropic.

Template: TChatInput

Input type for chat requests

Template: TEmbedInput

Input type for embedding requests

Template: TModerationInput

Input type for moderation requests

Hierarchy (View Summary)

BaseProvider
- AnthropicProvider

Implements

Constructors

constructor

new AnthropicProvider(): AnthropicProvider
Returns AnthropicProvider
Overrides BaseProvider.constructor
- Defined in src/providers/anthropic/AnthropicProvider.ts:78

Properties

`Protected`capabilities

capabilities: Partial<CapabilityMap> = {}

Support provider capabilities

`Protected`config

config: ProviderConnectionConfig | null = null

Current connection config

`Protected` `Optional`executors

executors?: Map<CapabilityKeyType, CapabilityExecutor<any, any, any>>

`Readonly`providerType

providerType: AIProviderType

Type of this provider (OpenAI, Anthropic, Gemini, etc.)

Methods

analyzeImage

analyzeImage(
    req: AIRequest<ClientImageAnalysisRequest>,
    executionContext: MultiModalExecutionContext,
    signal?: AbortSignal,
): Promise<AIResponse<NormalizedImageAnalysis[]>>
Execute an image analysis request
Parameters
- req: AIRequest<ClientImageAnalysisRequest>
  Unified AI request containing Image analysis input and options
- executionContext: MultiModalExecutionContext
  Execution context
- Optionalsignal: AbortSignal
  AbortSignal for request cancellation
Returns Promise<AIResponse<NormalizedImageAnalysis[]>>
AIResponse containing normalized image analysis
Implementation of ImageAnalysisCapability.analyzeImage
- Defined in src/providers/anthropic/AnthropicProvider.ts:226

analyzeImageStream

analyzeImageStream(
    req: AIRequest<ClientImageAnalysisRequest>,
    executionContext: MultiModalExecutionContext,
    signal?: AbortSignal,
): AsyncGenerator<AIResponseChunk<NormalizedImageAnalysis[]>>
Execute a streaming image analysis request
Parameters
- req: AIRequest<ClientImageAnalysisRequest>
  Unified AI request containing Image analysis input and options
- executionContext: MultiModalExecutionContext
  Execution context
- Optionalsignal: AbortSignal
  AbortSignal for request cancellation
Returns AsyncGenerator<AIResponseChunk<NormalizedImageAnalysis[]>>
AIResponseChunk containing normalized image analysis chunks
Implementation of ImageAnalysisStreamCapability.analyzeImageStream
- Defined in src/providers/anthropic/AnthropicProvider.ts:246

chat

chat(
    req: AIRequest<ClientChatRequest>,
    executionContext: MultiModalExecutionContext,
    signal?: AbortSignal,
): Promise<AIResponse<NormalizedChatMessage>>
Execute a non-streaming chat request.
Parameters
- req: AIRequest<ClientChatRequest>
  Unified AI request containing chat input and options
- executionContext: MultiModalExecutionContext
  Execution context
- Optionalsignal: AbortSignal
  AbortSignal for request cancellation
Returns Promise<AIResponse<NormalizedChatMessage>>
AIResponse containing generated text output
Implementation of ChatCapability.chat
- Defined in src/providers/anthropic/AnthropicProvider.ts:146

chatStream

chatStream(
    req: AIRequest<ClientChatRequest>,
    executionContext: MultiModalExecutionContext,
    signal?: AbortSignal,
): AsyncGenerator<AIResponseChunk<NormalizedChatMessage>>
Execute a streaming chat request.
Parameters
- req: AIRequest<ClientChatRequest>
  Unified AI request containing chat input and options
- executionContext: MultiModalExecutionContext
  Execution context
- Optionalsignal: AbortSignal
  AbortSignal for request cancellation
Returns AsyncGenerator<AIResponseChunk<NormalizedChatMessage>>
AsyncGenerator emitting streamed response chunks
Implementation of ChatStreamCapability.chatStream
- Defined in src/providers/anthropic/AnthropicProvider.ts:166

embed

embed(
    req: AIRequest<ClientEmbeddingRequest>,
    executionContext: MultiModalExecutionContext,
    signal?: AbortSignal,
): Promise<AIResponse<NormalizedEmbedding[]>>
Execute an embedding request.
Parameters
- req: AIRequest<ClientEmbeddingRequest>
  Unified AI request containing embedding input
- executionContext: MultiModalExecutionContext
  Execution context
- Optionalsignal: AbortSignal
  AbortSignal for request cancellation
Returns Promise<AIResponse<NormalizedEmbedding[]>>
AIResponse containing embedding vector(s)
Implementation of EmbedCapability.embed
- Defined in src/providers/anthropic/AnthropicProvider.ts:206

ensureInitialized

ensureInitialized(): void
Ensures that the provider has been initialized before use.

Returns void
Throws
Error if not initialized
Inherited from BaseProvider.ensureInitialized
- Defined in src/core/provider/BaseProvider.ts:351

getCapabilities

getCapabilities(): Partial<CapabilityMap>
Returns the currently registered capability implementations for this provider.

Returns Partial<CapabilityMap>
Registered capability map.
Inherited from BaseProvider.getCapabilities
- Defined in src/core/provider/BaseProvider.ts:120

getCapability

getCapability<C extends CapabilityKeyType>(
capability: C,
): C extends keyof CapabilityMap ? CapabilityMap[C] : ProviderCapability
Returns the implementation registered for a capability key.
Type Parameters
- C extends CapabilityKeyType
  Capability key type being resolved.
Parameters
- capability: C
  Capability key to resolve.
Returns C extends keyof CapabilityMap ? CapabilityMap[C] : ProviderCapability
Registered capability implementation.

Throws
When the capability is not registered.
Inherited from BaseProvider.getCapability
- Defined in src/core/provider/BaseProvider.ts:151

getMergedOptions

getMergedOptions(
capability: string,
runtimeOptions?: any,
): { generalParams: any; model: any; modelParams: any; providerParams: any }
Resolve and merge configuration for a capability.

Merge precedence (low → high):
1. providerDefaults (global provider-level defaults)
2. model capability defaults (models[modelName][capability])
3. runtimeOptions (request-level overrides)
Model resolution fallback: runtimeOptions.model → config.defaultModels[capability] → config.defaultModel
Parameters
- capability: string
  Name of the capability.
- runtimeOptions: any = {}
  Request-level override options.
Returns { generalParams: any; model: any; modelParams: any; providerParams: any }
Merged configuration with keys such as model, modelParams, providerParams, and generalParams.

Throws
When a model cannot be resolved for the requested capability.
Inherited from BaseProvider.getMergedOptions
- Defined in src/core/provider/BaseProvider.ts:288

getProviderType

getProviderType(): AIProviderType
Returns the canonical provider identifier.

Returns AIProviderType
Provider identifier.
Inherited from BaseProvider.getProviderType
- Defined in src/core/provider/BaseProvider.ts:111

hasCapability

hasCapability<C extends CapabilityKeyType>(capability: C): boolean
Type-safe runtime check for a capability. Allows safe casting after confirming the capability is registered.
Type Parameters
- C extends CapabilityKeyType
  Capability key.
Parameters
- capability: C
  Capability symbol.
Returns boolean
true when the capability is registered.
Inherited from BaseProvider.hasCapability
- Defined in src/core/provider/BaseProvider.ts:165

init

init(config: ProviderConnectionConfig): void
Initializes the Anthropic provider.
Parameters
- config: ProviderConnectionConfig
  Connection configuration for the provider
Returns void
Throws
Error if the API key is missing or invalid
Overrides BaseProvider.init
- Defined in src/providers/anthropic/AnthropicProvider.ts:88

isInitialized

isInitialized(): boolean
Check if the provider has been initialized.

Returns boolean
true when the provider has been initialized.
Inherited from BaseProvider.isInitialized
- Defined in src/core/provider/BaseProvider.ts:102

`Protected`mergeOptions

mergeOptions(...sources: any[]): any
Deep-merge multiple objects. Arrays override completely, objects are recursively merged, primitives override. Used for merging provider defaults, model configurations, and runtime options.
Parameters
- ...sources: any[]
  Objects to merge.
Returns any
Deep-merged object.
Inherited from BaseProvider.mergeOptions
- Defined in src/core/provider/BaseProvider.ts:208

moderation

moderation(
    req: AIRequest<ClientModerationRequest>,
    executionContext: MultiModalExecutionContext,
    signal?: AbortSignal,
): Promise<AIResponse<NormalizedModeration[]>>
Execute a moderation request.
Parameters
- req: AIRequest<ClientModerationRequest>
  Unified AI request containing moderation input
- executionContext: MultiModalExecutionContext
  Execution context
- Optionalsignal: AbortSignal
  AbortSignal for request cancellation
Returns Promise<AIResponse<NormalizedModeration[]>>
AIResponse containing moderation result(s)
Implementation of ModerationCapability.moderation
- Defined in src/providers/anthropic/AnthropicProvider.ts:186

ocr

ocr(
    req: AIRequest<ClientOCRRequest>,
    executionContext: MultiModalExecutionContext,
    signal?: AbortSignal,
): Promise<AIResponse<NormalizedOCRDocument[]>>
Execute a non-streaming OCR request.
Parameters
- req: AIRequest<ClientOCRRequest>
- executionContext: MultiModalExecutionContext
- Optionalsignal: AbortSignal
Returns Promise<AIResponse<NormalizedOCRDocument[]>>
Implementation of OCRCapability.ocr
- Defined in src/providers/anthropic/AnthropicProvider.ts:260

`Protected`registerCapability

registerCapability<C extends BuiltInCapabilityKey>(
capability: C,
impl: CapabilityMap[C],
): void
Register a capability implementation. Called by concrete providers to declare support for a capability.
Type Parameters
- C extends BuiltInCapabilityKey
  Built-in capability key.
Parameters
- capability: C
  Capability symbol.
- impl: CapabilityMap[C]
  Implementation of the capability.
Returns void
Inherited from BaseProvider.registerCapability
- Defined in src/core/provider/BaseProvider.ts:178
registerCapability(
capability: CustomCapabilityKey,
impl: ProviderCapability,
): void
Registers a custom capability implementation.
Parameters
- capability: CustomCapabilityKey
  Custom capability key.
- impl: ProviderCapability
  Capability implementation.
Returns void
Inherited from BaseProvider.registerCapability
- Defined in src/core/provider/BaseProvider.ts:185

setClientExecutors

setClientExecutors(
executors: Map<CapabilityKeyType, CapabilityExecutor<any, any, any>>,
): void
Attaches client-registered executors used for dispatching custom capabilities.
Parameters
- executors: Map<CapabilityKeyType, CapabilityExecutor<any, any, any>>
  Executor map.
Returns void
Inherited from BaseProvider.setClientExecutors
- Defined in src/core/provider/BaseProvider.ts:196

`Protected` `Static`sanitizeConstructorParams

sanitizeConstructorParams(
params: Record<string, unknown>,
): Record<string, unknown>
Strips keys from providerParams that must never be forwarded to an SDK constructor because they can redirect API traffic or override credentials.

Blocked keys across all supported SDKs:
- baseURL / baseUrl — redirects all provider traffic to an arbitrary host, causing the API key to be sent to an attacker-controlled server.
- apiKey / authToken — would silently override the key already set from config.
- httpAgent — custom HTTP agent can route traffic through an arbitrary proxy.
- fetch / globalFetch — custom fetch implementations can intercept all requests.
- httpOptions — Gemini SDK wrapper for baseUrl and custom fetch.
Parameters
- params: Record<string, unknown>
  Raw providerParams from caller config.
Returns Record<string, unknown>
Params with dangerous constructor keys removed.
Inherited from BaseProvider.sanitizeConstructorParams
- Defined in src/core/provider/BaseProvider.ts:83

Class AnthropicProvider

Template: TChatInput

Template: TEmbedInput

Template: TModerationInput

Hierarchy (View Summary)

Implements

Index

Constructors

Properties

Methods

Constructors

constructor

Returns AnthropicProvider

Properties

Protectedcapabilities

Protectedconfig

Protected Optionalexecutors

ReadonlyproviderType

Methods

analyzeImage

Parameters

Returns Promise<AIResponse<NormalizedImageAnalysis[]>>

analyzeImageStream

Parameters

Returns AsyncGenerator<AIResponseChunk<NormalizedImageAnalysis[]>>

chat

Parameters

Returns Promise<AIResponse<NormalizedChatMessage>>

chatStream

Parameters

Returns AsyncGenerator<AIResponseChunk<NormalizedChatMessage>>

embed

Parameters

Returns Promise<AIResponse<NormalizedEmbedding[]>>

ensureInitialized

Returns void

Throws

getCapabilities

Returns Partial<CapabilityMap>

getCapability

Type Parameters

Parameters

Returns C extends keyof CapabilityMap ? CapabilityMap[C] : ProviderCapability

Throws

getMergedOptions

Parameters

Returns { generalParams: any; model: any; modelParams: any; providerParams: any }

Throws

getProviderType

Returns AIProviderType

hasCapability

Type Parameters

Parameters

Returns boolean

init

Parameters

Returns void

Throws

isInitialized

Returns boolean

ProtectedmergeOptions

Parameters

Returns any

moderation

Parameters

Returns Promise<AIResponse<NormalizedModeration[]>>

ocr

Parameters

Returns Promise<AIResponse<NormalizedOCRDocument[]>>

ProtectedregisterCapability

Type Parameters

Parameters

Returns void

Parameters

Returns void

setClientExecutors

Parameters

Returns void

Protected StaticsanitizeConstructorParams

Parameters

`Protected`capabilities

`Protected`config

`Protected` `Optional`executors

`Readonly`providerType

`Protected`mergeOptions

`Protected`registerCapability

`Protected` `Static`sanitizeConstructorParams