@sprucelabs/sprucebot-llm 16.0.1 → 18.0.0

package/README.md

@@ -22,6 +22,9 @@ A TypeScript library for leveraging large language models to do... anything!
     * [Anthropic](#anthropic-adapter) - Claude models with prompt caching support
     * [Ollama](#ollama-adapter) - Run local models like Llama, Mistral, etc.
     * [Custom adapters](#custom-adapters) - Implement your own
+* [Track token usage](#tracking-token-usage)
+    * Read cumulative input/output/total tokens from any adapter
+    * Includes Anthropic prompt-cache token counts
 * Fully typed
 	* Built in modern TypeScript
 	* Fully typed schema-based state management (powered by `@sprucelabs/schema`)
@@ -192,6 +195,7 @@ adapter.setReasoningEffort('low')
 - `adapter.setModel(model)`: set a default model for all requests unless a Skill overrides it.
 - `adapter.setMessageMemoryLimit(limit)`: limit how many tracked messages are sent to OpenAI.
 - `adapter.setReasoningEffort(effort)`: set `reasoning_effort` for models that support it.
+- `adapter.getTokenUsage()`: returns cumulative [token usage](#tracking-token-usage) for this adapter. The OpenAI adapter currently returns zeros (not yet implemented).
 - `OpenAiAdapter.OpenAI`: assign a custom OpenAI client class (useful for tests).
 
 Requests are sent via `openai.chat.completions.create(...)` with messages built by the adapter from the Bot state and history.
@@ -234,6 +238,8 @@ Token usage (including cache creation and cache read tokens) is logged at the `i
 [TOKEN USAGE] input=1234 cache_create=800 cache_read=400 output=256
 ```
 
+The same numbers are also available programmatically via [`adapter.getTokenUsage()`](#tracking-token-usage).
+
 No configuration is required — caching is applied automatically.
 
 ### Ollama adapter
@@ -270,6 +276,48 @@ await bot.sendMessage('Hello!')
 
 The Ollama adapter connects to `http://localhost:11434/v1` by default (Ollama's OpenAI-compatible endpoint).
 
+### Tracking token usage
+
+Every adapter implements `getTokenUsage()`, which returns the **cumulative** token usage for that adapter instance since it was created:
+
+```ts
+import { LmmTokenUsage } from '@sprucelabs/sprucebot-llm'
+
+await bot.sendMessage('Hello!')
+
+const usage: LmmTokenUsage = adapter.getTokenUsage()
+// {
+//   inputTokens: 1234,
+//   outputTokens: 256,
+//   totalTokens: 1490,
+//   cacheCreationTokens: 800, // Anthropic only
+//   cacheReadTokens: 400,     // Anthropic only
+// }
+```
+
+The `LmmTokenUsage` shape:
+
+| Field | Type | Notes |
+|-------|------|-------|
+| `inputTokens` | `number` | Prompt / input tokens |
+| `outputTokens` | `number` | Completion / output tokens |
+| `totalTokens` | `number` | `inputTokens + outputTokens` |
+| `cacheCreationTokens` | `number` (optional) | Anthropic only — tokens written to the prompt cache |
+| `cacheReadTokens` | `number` (optional) | Anthropic only — tokens served from the prompt cache |
+
+**Adapter support:**
+
+| Adapter | `getTokenUsage()` |
+|---------|-------------------|
+| Anthropic | ✅ Fully tracked, including cache tokens |
+| OpenAI | ⚠️ Returns zeros — not yet implemented |
+| Ollama | ⚠️ Returns zeros — not yet implemented |
+
+**Things to know:**
+
+- **Cumulative, not per-call.** Totals accumulate across every `sendMessage` call on the adapter, and there is no reset method. To measure a single call, read `getTokenUsage()` before and after and diff the values.
+- **Per adapter, not per bot.** Usage lives on the adapter instance. If you share one adapter across multiple bots (e.g. through a single `SprucebotLlmFactory`), the totals aggregate across all of them.
+
 ### Custom adapters
 
 You can bring your own adapter by implementing the `LlmAdapter` interface and passing it to `SprucebotLlmFactory.Factory(...)`:
@@ -277,6 +325,8 @@ You can bring your own adapter by implementing the `LlmAdapter` interface and pa
 ```ts
 import {
 	LlmAdapter,
+	LllmReasoningEffort,
+	LmmTokenUsage,
 	SprucebotLlmBot,
 	SprucebotLlmFactory,
 } from '@sprucelabs/sprucebot-llm'
@@ -288,6 +338,15 @@ class MyAdapter implements LlmAdapter {
 		// Send to your model and return the model response as a string
 		return `echo: ${messages[messages.length - 1]?.message ?? ''}`
 	}
+
+	setModel(_model: string) {}
+	setReasoningEffort(_effort: LllmReasoningEffort) {}
+	setMemoryLimit(_limit: number) {}
+
+	// Return cumulative token usage for this adapter (zeros if you don't track it)
+	getTokenUsage(): LmmTokenUsage {
+		return { inputTokens: 0, outputTokens: 0, totalTokens: 0 }
+	}
 }
 
 const bots = SprucebotLlmFactory.Factory(new MyAdapter())

package/build/.spruce/settings.json

@@ -11,7 +11,7 @@
     "error"
   ],
   "regressionproof": {
-    "enabled": false,
+    "enabled": true,
     "projectName": "sprucebot-llm"
   }
 }

package/build/bots/adapters/AnthropicAdapter.d.ts

@@ -1,7 +1,7 @@
 import { Schema } from '@sprucelabs/schema';
 import { Log } from '@sprucelabs/spruce-skill-utils';
 import Anthropic from '@anthropic-ai/sdk';
-import { LlmAdapter, SprucebotLlmBot, SendMessageOptions, LllmReasoningEffort } from '../../llm.types';
+import { LlmAdapter, SprucebotLlmBot, SendMessageOptions, LllmReasoningEffort, LmmTokenUsage } from '../../llm.types';
 export default class AnthropicAdapter implements LlmAdapter {
     static Class?: new (apiKey: string, options?: AnthropicAdapterOptions) => LlmAdapter;
     static Anthropic: typeof Anthropic;
@@ -12,6 +12,7 @@ export default class AnthropicAdapter implements LlmAdapter {
     private memoryLimit?;
     private isThinkingEnabled;
     private log?;
+    private usage;
     private constructor();
     static Adapter(apiKey: string, options: AnthropicAdapterOptions): LlmAdapter;
     sendMessage(bot: SprucebotLlmBot<Schema>, options?: SendMessageOptions): Promise<string>;
@@ -19,6 +20,7 @@ export default class AnthropicAdapter implements LlmAdapter {
     setModel(model: string): void;
     setReasoningEffort(effort: LllmReasoningEffort): void;
     setMemoryLimit(limit: number): void;
+    getTokenUsage(): LmmTokenUsage;
 }
 export interface AnthropicAdapterOptions {
     log?: Log;

package/build/bots/adapters/AnthropicAdapter.js

@@ -10,6 +10,13 @@ class AnthropicAdapter {
     constructor(apiKey, options) {
         this.model = 'claude-sonnet-4-5';
         this.isThinkingEnabled = false;
+        this.usage = {
+            inputTokens: 0,
+            outputTokens: 0,
+            totalTokens: 0,
+            cacheCreationTokens: 0,
+            cacheReadTokens: 0,
+        };
         (0, schema_1.assertOptions)({ apiKey, maxTokens: options?.maxTokens }, [
             'apiKey',
             'maxTokens',
@@ -68,6 +75,17 @@ class AnthropicAdapter {
         }, sendOptions);
         const { usage } = response;
         this.log?.info(`[TOKEN USAGE] input=${usage.input_tokens} cache_create=${usage.cache_creation_input_tokens ?? 0} cache_read=${usage.cache_read_input_tokens ?? 0} output=${usage.output_tokens}`);
+        this.usage = {
+            inputTokens: this.usage.inputTokens + usage.input_tokens,
+            outputTokens: this.usage.outputTokens + usage.output_tokens,
+            totalTokens: this.usage.totalTokens +
+                usage.input_tokens +
+                usage.output_tokens,
+            cacheCreationTokens: (this.usage.cacheCreationTokens ?? 0) +
+                (usage.cache_creation_input_tokens ?? 0),
+            cacheReadTokens: (this.usage.cacheReadTokens ?? 0) +
+                (usage.cache_read_input_tokens ?? 0),
+        };
         const text = response.content
             .filter((block) => block.type === 'text')
             ?.map((block) => block.text)
@@ -84,6 +102,9 @@ class AnthropicAdapter {
     setMemoryLimit(limit) {
         this.memoryLimit = limit;
     }
+    getTokenUsage() {
+        return this.usage;
+    }
 }
 AnthropicAdapter.Anthropic = sdk_1.default;
 exports.default = AnthropicAdapter;

package/build/bots/adapters/OllamaAdapter.d.ts

@@ -1,6 +1,6 @@
 import { Schema } from '@sprucelabs/schema';
 import { Log } from '@sprucelabs/spruce-skill-utils';
-import { LlmAdapter, SprucebotLlmBot, SendMessageOptions, LllmReasoningEffort } from '../../llm.types';
+import { LlmAdapter, SprucebotLlmBot, SendMessageOptions, LllmReasoningEffort, LmmTokenUsage } from '../../llm.types';
 export default class OllamaAdapter implements LlmAdapter {
     static Class?: new (apiKey: string, options?: OllamaAdapterOptions) => LlmAdapter;
     private openai;
@@ -11,6 +11,7 @@ export default class OllamaAdapter implements LlmAdapter {
     setModel(model: string): void;
     setReasoningEffort(effort: LllmReasoningEffort): void;
     setMemoryLimit(_limit: number): void;
+    getTokenUsage(): LmmTokenUsage;
 }
 export interface OllamaAdapterOptions {
     log?: Log;

package/build/bots/adapters/OllamaAdapter.js

@@ -28,5 +28,12 @@ class OllamaAdapter {
     setMemoryLimit(_limit) {
         throw new Error('Method not implemented.');
     }
+    getTokenUsage() {
+        return {
+            inputTokens: 0,
+            outputTokens: 0,
+            totalTokens: 0,
+        };
+    }
 }
 exports.default = OllamaAdapter;

package/build/bots/adapters/OpenAiAdapter.d.ts

@@ -1,7 +1,7 @@
 import { Log } from '@sprucelabs/spruce-skill-utils';
 import OpenAI from 'openai';
 import { ReasoningEffort } from 'openai/resources';
-import { LlmAdapter, SendMessageOptions, SprucebotLlmBot } from '../../llm.types';
+import { LlmAdapter, LmmTokenUsage, SendMessageOptions, SprucebotLlmBot } from '../../llm.types';
 export default class OpenAiAdapter implements LlmAdapter {
     static Class?: new (apiKey: string, options?: OpenAiAdapterOptions) => LlmAdapter;
     static OpenAI: typeof OpenAI;
@@ -18,6 +18,7 @@ export default class OpenAiAdapter implements LlmAdapter {
     setModel(model: string): void;
     setMemoryLimit(limit: number): void;
     setReasoningEffort(effort: ReasoningEffort): void;
+    getTokenUsage(): LmmTokenUsage;
 }
 export declare const MESSAGE_RESPONSE_ERROR_MESSAGE = "Oh no! Something went wrong and I can't talk right now!";
 export interface OpenAiAdapterOptions {

package/build/bots/adapters/OpenAiAdapter.js

@@ -48,6 +48,13 @@ class OpenAiAdapter {
     setReasoningEffort(effort) {
         this.reasoningEffort = effort;
     }
+    getTokenUsage() {
+        return {
+            inputTokens: 0,
+            outputTokens: 0,
+            totalTokens: 0,
+        };
+    }
 }
 OpenAiAdapter.OpenAI = openai_1.default;
 exports.default = OpenAiAdapter;

package/build/esm/bots/adapters/AnthropicAdapter.d.ts

@@ -1,7 +1,7 @@
 import { Schema } from '@sprucelabs/schema';
 import { Log } from '@sprucelabs/spruce-skill-utils';
 import Anthropic from '@anthropic-ai/sdk';
-import { LlmAdapter, SprucebotLlmBot, SendMessageOptions, LllmReasoningEffort } from '../../llm.types';
+import { LlmAdapter, SprucebotLlmBot, SendMessageOptions, LllmReasoningEffort, LmmTokenUsage } from '../../llm.types';
 export default class AnthropicAdapter implements LlmAdapter {
     static Class?: new (apiKey: string, options?: AnthropicAdapterOptions) => LlmAdapter;
     static Anthropic: typeof Anthropic;
@@ -12,6 +12,7 @@ export default class AnthropicAdapter implements LlmAdapter {
     private memoryLimit?;
     private isThinkingEnabled;
     private log?;
+    private usage;
     private constructor();
     static Adapter(apiKey: string, options: AnthropicAdapterOptions): LlmAdapter;
     sendMessage(bot: SprucebotLlmBot<Schema>, options?: SendMessageOptions): Promise<string>;
@@ -19,6 +20,7 @@ export default class AnthropicAdapter implements LlmAdapter {
     setModel(model: string): void;
     setReasoningEffort(effort: LllmReasoningEffort): void;
     setMemoryLimit(limit: number): void;
+    getTokenUsage(): LmmTokenUsage;
 }
 export interface AnthropicAdapterOptions {
     log?: Log;

package/build/esm/bots/adapters/AnthropicAdapter.js

@@ -14,6 +14,13 @@ class AnthropicAdapter {
     constructor(apiKey, options) {
         this.model = 'claude-sonnet-4-5';
         this.isThinkingEnabled = false;
+        this.usage = {
+            inputTokens: 0,
+            outputTokens: 0,
+            totalTokens: 0,
+            cacheCreationTokens: 0,
+            cacheReadTokens: 0,
+        };
         assertOptions({ apiKey, maxTokens: options === null || options === void 0 ? void 0 : options.maxTokens }, [
             'apiKey',
             'maxTokens',
@@ -39,7 +46,7 @@ class AnthropicAdapter {
     }
     sendHandler(params, sendOptions) {
         return __awaiter(this, void 0, void 0, function* () {
-            var _a, _b, _c, _d;
+            var _a, _b, _c, _d, _e, _f, _g, _h;
             const { messages: openAiMessages, model } = params;
             const messages = [];
             const cacheMarkerIdx = openAiMessages.findIndex((msg) => msg.cache_marker);
@@ -73,8 +80,19 @@ class AnthropicAdapter {
             }, sendOptions);
             const { usage } = response;
             (_a = this.log) === null || _a === void 0 ? void 0 : _a.info(`[TOKEN USAGE] input=${usage.input_tokens} cache_create=${(_b = usage.cache_creation_input_tokens) !== null && _b !== void 0 ? _b : 0} cache_read=${(_c = usage.cache_read_input_tokens) !== null && _c !== void 0 ? _c : 0} output=${usage.output_tokens}`);
-            const text = (_d = response.content
-                .filter((block) => block.type === 'text')) === null || _d === void 0 ? void 0 : _d.map((block) => block.text).join('').trim();
+            this.usage = {
+                inputTokens: this.usage.inputTokens + usage.input_tokens,
+                outputTokens: this.usage.outputTokens + usage.output_tokens,
+                totalTokens: this.usage.totalTokens +
+                    usage.input_tokens +
+                    usage.output_tokens,
+                cacheCreationTokens: ((_d = this.usage.cacheCreationTokens) !== null && _d !== void 0 ? _d : 0) +
+                    ((_e = usage.cache_creation_input_tokens) !== null && _e !== void 0 ? _e : 0),
+                cacheReadTokens: ((_f = this.usage.cacheReadTokens) !== null && _f !== void 0 ? _f : 0) +
+                    ((_g = usage.cache_read_input_tokens) !== null && _g !== void 0 ? _g : 0),
+            };
+            const text = (_h = response.content
+                .filter((block) => block.type === 'text')) === null || _h === void 0 ? void 0 : _h.map((block) => block.text).join('').trim();
             return text;
         });
     }
@@ -87,6 +105,9 @@ class AnthropicAdapter {
     setMemoryLimit(limit) {
         this.memoryLimit = limit;
     }
+    getTokenUsage() {
+        return this.usage;
+    }
 }
 AnthropicAdapter.Anthropic = Anthropic;
 export default AnthropicAdapter;

package/build/esm/bots/adapters/OllamaAdapter.d.ts

@@ -1,6 +1,6 @@
 import { Schema } from '@sprucelabs/schema';
 import { Log } from '@sprucelabs/spruce-skill-utils';
-import { LlmAdapter, SprucebotLlmBot, SendMessageOptions, LllmReasoningEffort } from '../../llm.types';
+import { LlmAdapter, SprucebotLlmBot, SendMessageOptions, LllmReasoningEffort, LmmTokenUsage } from '../../llm.types';
 export default class OllamaAdapter implements LlmAdapter {
     static Class?: new (apiKey: string, options?: OllamaAdapterOptions) => LlmAdapter;
     private openai;
@@ -11,6 +11,7 @@ export default class OllamaAdapter implements LlmAdapter {
     setModel(model: string): void;
     setReasoningEffort(effort: LllmReasoningEffort): void;
     setMemoryLimit(_limit: number): void;
+    getTokenUsage(): LmmTokenUsage;
 }
 export interface OllamaAdapterOptions {
     log?: Log;

package/build/esm/bots/adapters/OllamaAdapter.js

@@ -33,4 +33,11 @@ export default class OllamaAdapter {
     setMemoryLimit(_limit) {
         throw new Error('Method not implemented.');
     }
+    getTokenUsage() {
+        return {
+            inputTokens: 0,
+            outputTokens: 0,
+            totalTokens: 0,
+        };
+    }
 }

package/build/esm/bots/adapters/OpenAiAdapter.d.ts

@@ -1,7 +1,7 @@
 import { Log } from '@sprucelabs/spruce-skill-utils';
 import OpenAI from 'openai';
 import { ReasoningEffort } from 'openai/resources';
-import { LlmAdapter, SendMessageOptions, SprucebotLlmBot } from '../../llm.types';
+import { LlmAdapter, LmmTokenUsage, SendMessageOptions, SprucebotLlmBot } from '../../llm.types';
 export default class OpenAiAdapter implements LlmAdapter {
     static Class?: new (apiKey: string, options?: OpenAiAdapterOptions) => LlmAdapter;
     static OpenAI: typeof OpenAI;
@@ -18,6 +18,7 @@ export default class OpenAiAdapter implements LlmAdapter {
     setModel(model: string): void;
     setMemoryLimit(limit: number): void;
     setReasoningEffort(effort: ReasoningEffort): void;
+    getTokenUsage(): LmmTokenUsage;
 }
 export declare const MESSAGE_RESPONSE_ERROR_MESSAGE = "Oh no! Something went wrong and I can't talk right now!";
 export interface OpenAiAdapterOptions {

package/build/esm/bots/adapters/OpenAiAdapter.js

@@ -53,6 +53,13 @@ class OpenAiAdapter {
     setReasoningEffort(effort) {
         this.reasoningEffort = effort;
     }
+    getTokenUsage() {
+        return {
+            inputTokens: 0,
+            outputTokens: 0,
+            totalTokens: 0,
+        };
+    }
 }
 OpenAiAdapter.OpenAI = OpenAI;
 export default OpenAiAdapter;

package/build/esm/index.d.ts

@@ -10,7 +10,7 @@ export { default as SpyLlmAdapter } from './tests/SpyAdapter';
 export { default as SpyOpenAiApi } from './bots/adapters/SpyOpenAiModule';
 export { default as OllamaAdapter } from './bots/adapters/OllamaAdapter';
 export * from './bots/adapters/OllamaAdapter';
-export { default as AthropicAdapter } from './bots/adapters/AnthropicAdapter';
+export { default as AnthropicAdapter } from './bots/adapters/AnthropicAdapter';
 export * from './bots/adapters/AnthropicAdapter';
 export { default as LlmAdapterLoader } from './bots/adapters/LlmAdapterLoader';
 export * from './bots/adapters/LlmAdapterLoader';

package/build/esm/index.js

@@ -10,7 +10,7 @@ export { default as SpyLlmAdapter } from './tests/SpyAdapter.js';
 export { default as SpyOpenAiApi } from './bots/adapters/SpyOpenAiModule.js';
 export { default as OllamaAdapter } from './bots/adapters/OllamaAdapter.js';
 export * from './bots/adapters/OllamaAdapter.js';
-export { default as AthropicAdapter } from './bots/adapters/AnthropicAdapter.js';
+export { default as AnthropicAdapter } from './bots/adapters/AnthropicAdapter.js';
 export * from './bots/adapters/AnthropicAdapter.js';
 export { default as LlmAdapterLoader } from './bots/adapters/LlmAdapterLoader.js';
 export * from './bots/adapters/LlmAdapterLoader.js';

package/build/esm/llm.types.d.ts

@@ -21,6 +21,14 @@ export interface LlmAdapter {
     setModel(model: string): void;
     setReasoningEffort(effort: LllmReasoningEffort): void;
     setMemoryLimit(limit: number): void;
+    getTokenUsage(): LmmTokenUsage;
+}
+export interface LmmTokenUsage {
+    inputTokens: number;
+    outputTokens: number;
+    totalTokens: number;
+    cacheCreationTokens?: number;
+    cacheReadTokens?: number;
 }
 export interface PromptOptions<StateSchema extends Schema, State extends SchemaValues<StateSchema> = SchemaValues<StateSchema>> {
     /**

package/build/esm/tests/SpyAdapter.d.ts

@@ -1,5 +1,5 @@
 import { OpenAiAdapterOptions } from '../bots/adapters/OpenAiAdapter';
-import { LllmReasoningEffort, LlmAdapter, SendMessageOptions, SprucebotLlmBot } from '../llm.types';
+import { LllmReasoningEffort, LlmAdapter, LmmTokenUsage, SendMessageOptions, SprucebotLlmBot } from '../llm.types';
 export default class SpyLlmAdapter implements LlmAdapter {
     static instance: SpyLlmAdapter;
     lastSendMessageBot?: SprucebotLlmBot;
@@ -16,4 +16,5 @@ export default class SpyLlmAdapter implements LlmAdapter {
     setModel(model: string): void;
     setReasoningEffort(_effort: LllmReasoningEffort): void;
     setMemoryLimit(_limit: number): void;
+    getTokenUsage(): LmmTokenUsage;
 }

package/build/esm/tests/SpyAdapter.js

@@ -34,4 +34,11 @@ export default class SpyLlmAdapter {
     }
     setReasoningEffort(_effort) { }
     setMemoryLimit(_limit) { }
+    getTokenUsage() {
+        return {
+            inputTokens: 0,
+            outputTokens: 0,
+            totalTokens: 0,
+        };
+    }
 }

package/build/index.d.ts

@@ -10,7 +10,7 @@ export { default as SpyLlmAdapter } from './tests/SpyAdapter';
 export { default as SpyOpenAiApi } from './bots/adapters/SpyOpenAiModule';
 export { default as OllamaAdapter } from './bots/adapters/OllamaAdapter';
 export * from './bots/adapters/OllamaAdapter';
-export { default as AthropicAdapter } from './bots/adapters/AnthropicAdapter';
+export { default as AnthropicAdapter } from './bots/adapters/AnthropicAdapter';
 export * from './bots/adapters/AnthropicAdapter';
 export { default as LlmAdapterLoader } from './bots/adapters/LlmAdapterLoader';
 export * from './bots/adapters/LlmAdapterLoader';

package/build/index.js

@@ -17,7 +17,7 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
     return (mod && mod.__esModule) ? mod : { "default": mod };
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.MockAdapterLoader = exports.LlmAdapterLoader = exports.AthropicAdapter = exports.OllamaAdapter = exports.SpyOpenAiApi = exports.SpyLlmAdapter = exports.SpyLllmBot = exports.MockLlmSkill = exports.SprucebotLlmError = exports.OpenAiAdapter = exports.SprucebotLlmSkillImpl = exports.SprucebotLlmBotImpl = exports.SprucebotLlmFactory = void 0;
+exports.MockAdapterLoader = exports.LlmAdapterLoader = exports.AnthropicAdapter = exports.OllamaAdapter = exports.SpyOpenAiApi = exports.SpyLlmAdapter = exports.SpyLllmBot = exports.MockLlmSkill = exports.SprucebotLlmError = exports.OpenAiAdapter = exports.SprucebotLlmSkillImpl = exports.SprucebotLlmBotImpl = exports.SprucebotLlmFactory = void 0;
 var SprucebotLlmFactory_1 = require("./bots/SprucebotLlmFactory");
 Object.defineProperty(exports, "SprucebotLlmFactory", { enumerable: true, get: function () { return __importDefault(SprucebotLlmFactory_1).default; } });
 var SprucebotLlmBotImpl_1 = require("./bots/SprucebotLlmBotImpl");
@@ -41,7 +41,7 @@ var OllamaAdapter_1 = require("./bots/adapters/OllamaAdapter");
 Object.defineProperty(exports, "OllamaAdapter", { enumerable: true, get: function () { return __importDefault(OllamaAdapter_1).default; } });
 __exportStar(require("./bots/adapters/OllamaAdapter"), exports);
 var AnthropicAdapter_1 = require("./bots/adapters/AnthropicAdapter");
-Object.defineProperty(exports, "AthropicAdapter", { enumerable: true, get: function () { return __importDefault(AnthropicAdapter_1).default; } });
+Object.defineProperty(exports, "AnthropicAdapter", { enumerable: true, get: function () { return __importDefault(AnthropicAdapter_1).default; } });
 __exportStar(require("./bots/adapters/AnthropicAdapter"), exports);
 var LlmAdapterLoader_1 = require("./bots/adapters/LlmAdapterLoader");
 Object.defineProperty(exports, "LlmAdapterLoader", { enumerable: true, get: function () { return __importDefault(LlmAdapterLoader_1).default; } });

package/build/llm.types.d.ts

@@ -21,6 +21,14 @@ export interface LlmAdapter {
     setModel(model: string): void;
     setReasoningEffort(effort: LllmReasoningEffort): void;
     setMemoryLimit(limit: number): void;
+    getTokenUsage(): LmmTokenUsage;
+}
+export interface LmmTokenUsage {
+    inputTokens: number;
+    outputTokens: number;
+    totalTokens: number;
+    cacheCreationTokens?: number;
+    cacheReadTokens?: number;
 }
 export interface PromptOptions<StateSchema extends Schema, State extends SchemaValues<StateSchema> = SchemaValues<StateSchema>> {
     /**

package/build/tests/SpyAdapter.d.ts

@@ -1,5 +1,5 @@
 import { OpenAiAdapterOptions } from '../bots/adapters/OpenAiAdapter';
-import { LllmReasoningEffort, LlmAdapter, SendMessageOptions, SprucebotLlmBot } from '../llm.types';
+import { LllmReasoningEffort, LlmAdapter, LmmTokenUsage, SendMessageOptions, SprucebotLlmBot } from '../llm.types';
 export default class SpyLlmAdapter implements LlmAdapter {
     static instance: SpyLlmAdapter;
     lastSendMessageBot?: SprucebotLlmBot;
@@ -16,4 +16,5 @@ export default class SpyLlmAdapter implements LlmAdapter {
     setModel(model: string): void;
     setReasoningEffort(_effort: LllmReasoningEffort): void;
     setMemoryLimit(_limit: number): void;
+    getTokenUsage(): LmmTokenUsage;
 }

package/build/tests/SpyAdapter.js

@@ -25,5 +25,12 @@ class SpyLlmAdapter {
     }
     setReasoningEffort(_effort) { }
     setMemoryLimit(_limit) { }
+    getTokenUsage() {
+        return {
+            inputTokens: 0,
+            outputTokens: 0,
+            totalTokens: 0,
+        };
+    }
 }
 exports.default = SpyLlmAdapter;

package/package.json

@@ -8,7 +8,7 @@
       "eta"
     ]
   },
-  "version": "16.0.1",
+  "version": "18.0.0",
   "files": [
     "build"
   ],