@xberg-io/liter-llm 1.11.2 → 1.16.0

package/README.md

@@ -307,7 +307,7 @@ To use the MCP server inside a coding agent, install the **liter-llm plugin** fr
 
 ## Part of Xberg.io
 
-- [Xberg](https://github.com/xberg-io/xberg) — document intelligence: text, tables, metadata from 91+ formats with optional OCR.
+- [Xberg](https://github.com/xberg-io/xberg) — document intelligence: text, tables, metadata from 101 formats with optional OCR.
 - [Xberg Enterprise](https://github.com/xberg-io/xberg-enterprise) — managed extraction API with SDKs, dashboards, and observability.
 - [crawlberg](https://github.com/xberg-io/crawlberg) — web crawling and scraping with HTML→Markdown and headless-Chrome fallback.
 - [html-to-markdown](https://github.com/xberg-io/html-to-markdown) — fast, lossless HTML→Markdown engine.

package/index.d.ts

@@ -1,5 +1,5 @@
 // This file is auto-generated by alef — DO NOT EDIT.
-// alef:hash:645d8ff0f3a0ef7684f9b9ea4ce1ae251a78a2a3481ecdeabf7643c53d644bb0
+// alef:hash:938b8ffd93c3e30d74875f875188c728addee9d4b80a97a7ac22716b3154c7c5
 // To regenerate: alef generate
 // To verify freshness: alef verify --exit-code
 /* eslint-disable */
@@ -373,6 +373,27 @@ export declare enum BatchStatus {
   Cancelled = "cancelled",
 }
 
+/**
+ * AWS Bedrock configuration.
+ *
+ * All fields are optional; anything left unset falls back to the standard
+ * AWS environment variables (`AWS_DEFAULT_REGION` / `AWS_REGION`,
+ * `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY`, `AWS_SESSION_TOKEN`,
+ * `BEDROCK_CROSS_REGION`).
+ */
+export interface BedrockConfig {
+  /** AWS region (e.g. `"us-east-1"`). */
+  readonly region?: string
+  /** Cross-region inference profile prefix (e.g. `"us"`). */
+  readonly crossRegionPrefix?: string
+  /** Explicit AWS access key ID. */
+  readonly accessKeyId?: string
+  /** Explicit AWS secret access key. */
+  readonly secretAccessKey?: string
+  /** Explicit AWS session token (temporary credentials). */
+  readonly sessionToken?: string
+}
+
 /** Configuration for budget enforcement. */
 export interface BudgetConfig {
   /** Maximum total spend across all models, in USD.  `None` means unlimited. */
@@ -500,7 +521,7 @@ export interface ChatCompletionRequest {
   readonly streamOptions?: StreamOptions
   /** Random seed for reproducible outputs. Provider support varies. */
   readonly seed?: number
-  /** Reasoning effort level (low, medium, high) for extended-thinking models. */
+  /** Reasoning effort level (minimal, low, medium, high, max) for extended-thinking models. */
   readonly reasoningEffort?: ReasoningEffort
   /**
    * Output modalities to request from the model.
@@ -653,6 +674,12 @@ export interface CreateResponseRequest {
   readonly maxOutputTokens?: number
   /** Optional metadata. */
   readonly metadata?: JsonValue
+  /**
+   * Whether to stream the response.
+   *
+   * Managed by the client layer — do not set directly.
+   */
+  readonly stream?: boolean
 }
 
 /** Request to generate speech audio from text. */
@@ -1048,6 +1075,108 @@ export interface JsonSchemaFormat {
   readonly strict?: boolean
 }
 
+/** Budget enforcement configuration. */
+export interface LlmBudgetConfig {
+  /** Global spend limit in USD. */
+  readonly globalLimit?: number
+  /** Per-model spend limits in USD, keyed by model name. */
+  readonly modelLimits?: Record<string, number>
+  /** Enforcement mode: `"hard"` (reject over-budget requests) or `"soft"` (log only). */
+  readonly enforcement?: string
+}
+
+/** Response cache configuration. */
+export interface LlmCacheConfig {
+  /** Maximum number of cached entries. */
+  readonly maxEntries?: number
+  /** Cache entry time-to-live, in seconds. */
+  readonly ttlSeconds?: number
+  /** Cache backend name (e.g. `"memory"`, or an `opendal` scheme). */
+  readonly backend?: string
+  /** Backend-specific configuration key/value pairs. */
+  readonly backendConfig?: Record<string, string>
+}
+
+/**
+ * Canonical configuration for an LLM client.
+ *
+ * All fields except `model` are optional so that partially-specified
+ * configs (e.g. from environment-driven defaults) round-trip cleanly.
+ * Convert to a runtime client configuration via
+ * [`LlmConfig::into_client_builder`].
+ *
+ * `temperature` and `max_tokens` are request-time parameters rather than
+ * client-level settings; they are carried on this struct for callers to
+ * read when building individual requests, and are intentionally **not**
+ * mapped by [`LlmConfig::into_client_builder`].
+ */
+export interface LlmConfig {
+  /** Model identifier (e.g. `"gpt-4o"`, `"bedrock/anthropic.claude-3-sonnet-20240229-v1:0"`). */
+  readonly model?: string
+  /** API key for authentication. */
+  readonly apiKey?: string
+  /**
+   * Override base URL. When set, all requests go here and provider
+   * auto-detection is skipped.
+   */
+  readonly baseUrl?: string
+  /** Request timeout, in seconds. */
+  readonly timeoutSecs?: number
+  /** Maximum number of retries on 429 / 5xx responses. */
+  readonly maxRetries?: number
+  /** Sampling temperature for requests built from this config. */
+  readonly temperature?: number
+  /** Maximum number of tokens to generate for requests built from this config. */
+  readonly maxTokens?: number
+  /**
+   * Automatically load the API key from the provider's environment variable
+   * when no explicit key is provided (default: `true`).
+   */
+  readonly loadEnv?: boolean
+  /** Extra headers sent on every request. */
+  readonly headers?: Record<string, string>
+  /** Custom provider configurations, in addition to the built-in providers. */
+  readonly providers?: Array<LlmProviderConfig>
+  /** Response cache configuration. */
+  readonly cache?: LlmCacheConfig
+  /** Budget enforcement configuration. */
+  readonly budget?: LlmBudgetConfig
+  /** Per-model rate limiting configuration. */
+  readonly rateLimit?: LlmRateLimitConfig
+  /** Enable per-request cost tracking. */
+  readonly costTracking?: boolean
+  /** Enable OpenTelemetry-compatible tracing spans. */
+  readonly tracing?: boolean
+  /** Cooldown duration after transient errors, in seconds. */
+  readonly cooldownSecs?: number
+  /** Background health check interval, in seconds. */
+  readonly healthCheckSecs?: number
+  /** AWS Bedrock configuration (region, credentials, cross-region routing). */
+  readonly bedrock?: BedrockConfig
+}
+
+/** A custom provider configuration entry. */
+export interface LlmProviderConfig {
+  /** Provider name, used to key model prefix matching. */
+  readonly name?: string
+  /** Base URL for the provider's OpenAI-compatible API. */
+  readonly baseUrl?: string
+  /** Header name used to carry the API key (defaults to `Authorization` when unset). */
+  readonly authHeader?: string
+  /** Model name prefixes routed to this provider (e.g. `["my-provider/"]`). */
+  readonly modelPrefixes?: Array<string>
+}
+
+/** Per-model rate limiting configuration. */
+export interface LlmRateLimitConfig {
+  /** Requests per minute limit. */
+  readonly rpm?: number
+  /** Tokens per minute limit. */
+  readonly tpm?: number
+  /** Rate limit window, in seconds. */
+  readonly windowSeconds?: number
+}
+
 /** A chat message in a conversation. */
 export type Message =
   | { role: 'system'; 0: SystemMessage }
@@ -1430,6 +1559,8 @@ export declare enum ReasoningEffort {
   Low = "low",
   Medium = "medium",
   High = "high",
+  Minimal = "minimal",
+  Max = "max",
 }
 
 /** Result of a [`refresh_catalog`] call. */
@@ -1744,8 +1875,15 @@ export declare enum ToolChoiceMode {
 
 /** Tool execution result returned to the model. */
 export interface ToolMessage {
-  /** Result of the tool execution. */
-  readonly content?: string
+  /**
+   * Result of the tool execution as plain text or an array of content parts
+   * (text, images, documents, audio), mirroring [`UserMessage::content`].
+   *
+   * `#[serde(untagged)]` on [`UserContent`] means a bare JSON string still
+   * deserialises into `Text`, so tool results persisted before this field
+   * carried structured content continue to round-trip.
+   */
+  readonly content?: UserContent
   /** ID of the tool call this result responds to. */
   readonly toolCallId?: string
   /** Optional tool/function name. */

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xberg-io/liter-llm",
-  "version": "1.11.2",
+  "version": "1.16.0",
   "description": "Universal LLM API client with Rust-powered polyglot bindings.",
   "license": "MIT",
   "repository": {
@@ -18,12 +18,12 @@
   },
   "files": ["index.js", "index.d.ts", "*.node"],
   "optionalDependencies": {
-    "@xberg-io/liter-llm-linux-x64-gnu": "1.11.2",
-    "@xberg-io/liter-llm-linux-arm64-gnu": "1.11.2",
-    "@xberg-io/liter-llm-darwin-x64": "1.11.2",
-    "@xberg-io/liter-llm-darwin-arm64": "1.11.2",
-    "@xberg-io/liter-llm-win32-x64-msvc": "1.11.2",
-    "@xberg-io/liter-llm-win32-arm64-msvc": "1.11.2"
+    "@xberg-io/liter-llm-linux-x64-gnu": "1.16.0",
+    "@xberg-io/liter-llm-linux-arm64-gnu": "1.16.0",
+    "@xberg-io/liter-llm-darwin-x64": "1.16.0",
+    "@xberg-io/liter-llm-darwin-arm64": "1.16.0",
+    "@xberg-io/liter-llm-win32-x64-msvc": "1.16.0",
+    "@xberg-io/liter-llm-win32-arm64-msvc": "1.16.0"
   },
   "napi": {
     "packageName": "@xberg-io/liter-llm",