agentic-api 2.0.314 → 2.0.585

package/dist/src/llm/pricing.js

@@ -0,0 +1,129 @@
+"use strict";
+/**
+ * LLM Pricing
+ *
+ * Gère le pricing des modèles.
+ * Les mappings de modèles sont dans openai.ts et xai.ts
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.LLMxai = exports.LLMopenai = exports.modelPricing = void 0;
+exports.calculateCost = calculateCost;
+exports.accumulateCost = accumulateCost;
+exports.LLM = LLM;
+const config_1 = require("./config");
+const openai_1 = require("./openai");
+const xai_1 = require("./xai");
+//
+// Pricing par modèle
+// - https://platform.openai.com/docs/pricing#latest-models
+// - https://x.ai/api (pour xAI)
+exports.modelPricing = {
+    // OpenAI GPT-4 family
+    "gpt-4.5-preview": { input: 0.000075, cachedInput: 0.0000325, output: 0.000125 },
+    "gpt-4.1": { input: 0.000002, cachedInput: 0.0000005, output: 0.000008 },
+    "gpt-4.1-mini": { input: 0.0000004, cachedInput: 0.0000001, output: 0.0000016 },
+    "gpt-4.1-nano": { input: 0.0000001, cachedInput: 0.000000025, output: 0.0000004 },
+    "gpt-4o": { input: 0.0000025, cachedInput: 0.00000125, output: 0.00001 },
+    "gpt-4o-audio-preview": { input: 0.0000025, output: 0.00001 },
+    "gpt-4o-realtime-preview": { input: 0.000005, cachedInput: 0.0000025, output: 0.00002 },
+    "gpt-4o-search-preview": { input: 0.000005, cachedInput: 0.0000025, output: 0.00002 },
+    "gpt-4o-mini": { input: 0.00000015, cachedInput: 0.000000075, output: 0.0000006 },
+    "gpt-4o-mini-audio-preview": { input: 0.00000015, output: 0.0000006 },
+    "gpt-4o-mini-realtime-preview": { input: 0.0000006, cachedInput: 0.0000003, output: 0.0000024 },
+    "gpt-4o-mini-search-preview": { input: 0.0000015, cachedInput: 0.00000075, output: 0.000006 },
+    "gpt-4o-mini-transcribe": { input: 0.0000015, output: 0.000006 },
+    // OpenAI GPT-5 family
+    "gpt-5": { input: 0.00000125, output: 0.00001 },
+    "gpt-5.1": { input: 0.00000125, output: 0.00001 },
+    "gpt-5.2": { input: 0.00000175, cachedInput: 0.000000175, output: 0.000014 },
+    "gpt-5-mini": { input: 0.00000025, output: 0.000002 },
+    "gpt-5-nano": { input: 0.00000005, output: 0.0000004 },
+    // OpenAI o-series
+    "o1": { input: 0.000015, cachedInput: 0.0000075, output: 0.00006 },
+    "o4-mini": { input: 0.0000011, cachedInput: 0.00000055, output: 0.0000044 },
+    "o3-mini": { input: 0.0000011, cachedInput: 0.00000055, output: 0.0000044 },
+    "o1-mini": { input: 0.0000011, cachedInput: 0.00000055, output: 0.0000044 },
+    // OpenAI Embeddings
+    "text-embedding-3-small": { input: 0.00000002, output: 0 },
+    "text-embedding-3-large": { input: 0.00000013, output: 0 },
+    // OpenAI Audio
+    "whisper-1": { input: 0.0001, output: 0 }, // per second
+    // xAI Grok 4 family (Décembre 2024)
+    // https://x.ai/api#pricing - Prix par token ($/1M tokens divisé par 1M)
+    "grok-4": { input: 0.000003, cachedInput: 0.000003, output: 0.000015 },
+    "grok-4-1-fast-reasoning": { input: 0.0000002, cachedInput: 0.0000002, output: 0.0000005 },
+    "grok-4-1-fast-non-reasoning": { input: 0.0000002, cachedInput: 0.0000002, output: 0.0000005 },
+    "grok-4-fast-reasoning": { input: 0.0000002, cachedInput: 0.0000002, output: 0.0000005 },
+    "grok-4-fast-non-reasoning": { input: 0.0000002, cachedInput: 0.0000002, output: 0.0000005 },
+    "grok-code-fast-1": { input: 0.0000002, output: 0.0000015 },
+};
+function calculateCost(model, usage) {
+    if (!usage) {
+        return 0;
+    }
+    if (!exports.modelPricing[model]) {
+        console.warn(`⚠️ Unknown model for pricing: ${model}`);
+        return 0;
+    }
+    const pricing = exports.modelPricing[model];
+    const cost = usage.prompt_tokens * pricing.input +
+        usage.completion_tokens * pricing.output;
+    return cost;
+}
+function accumulateCost(currentUsage, model, usage) {
+    if (!usage) {
+        return 0;
+    }
+    currentUsage.prompt += usage.prompt_tokens || 0;
+    currentUsage.completion += usage.completion_tokens || 0;
+    currentUsage.total += usage.total_tokens || 0;
+    const cost = calculateCost(model, usage);
+    currentUsage.cost += cost;
+    return currentUsage.cost;
+}
+//
+// Mapping provider → modèles
+const LLMmapping = {
+    openai: openai_1.LLMopenai,
+    xai: xai_1.LLMxai
+};
+/**
+ * Retourne le mapping des modèles selon le provider
+ *
+ * @param provider - Provider ('openai' | 'xai') ou instance OpenAI (legacy)
+ * @param forceThinking - Force reasoning_effort à high
+ */
+function LLM(provider, forceThinking) {
+    let resolvedProvider;
+    //
+    // Support legacy: instance OpenAI passée directement
+    if (provider && typeof provider === 'object' && provider.baseURL) {
+        resolvedProvider = provider.baseURL.includes('x.ai') ? 'xai' : 'openai';
+    }
+    else if (typeof provider === 'string') {
+        resolvedProvider = provider;
+    }
+    else {
+        resolvedProvider = (0, config_1.getDefaultProvider)();
+    }
+    const mapping = LLMmapping[resolvedProvider] || LLMmapping.openai;
+    //
+    // Clone pour éviter de muter l'original
+    const result = { ...mapping };
+    //
+    // Force reasoning_effort to high if thinking is enabled
+    if (forceThinking) {
+        Object.keys(result).forEach(key => {
+            if (result[key]?.reasoning_effort) {
+                result[key] = { ...result[key], reasoning_effort: "high" };
+            }
+        });
+    }
+    return result;
+}
+//
+// Re-export des mappings pour rétrocompatibilité
+var openai_2 = require("./openai");
+Object.defineProperty(exports, "LLMopenai", { enumerable: true, get: function () { return openai_2.LLMopenai; } });
+var xai_2 = require("./xai");
+Object.defineProperty(exports, "LLMxai", { enumerable: true, get: function () { return xai_2.LLMxai; } });

package/dist/src/llm/xai.d.ts

@@ -0,0 +1,17 @@
+/**
+ * xAI (Grok) Model Mappings
+ *
+ * Configuration des modèles pour le provider xAI
+ * Note: xAI ne supporte pas les embeddings ni Whisper nativement
+ *
+ * Modèles disponibles (Décembre 2024):
+ * - grok-4: Flagship model ($3.00/$15.00 per 1M tokens)
+ * - grok-4-1-fast-reasoning: Fast reasoning ($0.20/$0.50 per 1M tokens)
+ * - grok-4-1-fast-non-reasoning: Fast non-reasoning ($0.20/$0.50 per 1M tokens)
+ * - grok-4-fast-reasoning: Fast reasoning ($0.20/$0.50 per 1M tokens)
+ * - grok-4-fast-non-reasoning: Fast non-reasoning ($0.20/$0.50 per 1M tokens)
+ * - grok-code-fast-1: Agentic coding ($0.20/$1.50 per 1M tokens)
+ *
+ * Pricing: https://x.ai/api#pricing
+ */
+export declare const LLMxai: Record<string, any>;

package/dist/src/llm/xai.js

@@ -0,0 +1,90 @@
+"use strict";
+/**
+ * xAI (Grok) Model Mappings
+ *
+ * Configuration des modèles pour le provider xAI
+ * Note: xAI ne supporte pas les embeddings ni Whisper nativement
+ *
+ * Modèles disponibles (Décembre 2024):
+ * - grok-4: Flagship model ($3.00/$15.00 per 1M tokens)
+ * - grok-4-1-fast-reasoning: Fast reasoning ($0.20/$0.50 per 1M tokens)
+ * - grok-4-1-fast-non-reasoning: Fast non-reasoning ($0.20/$0.50 per 1M tokens)
+ * - grok-4-fast-reasoning: Fast reasoning ($0.20/$0.50 per 1M tokens)
+ * - grok-4-fast-non-reasoning: Fast non-reasoning ($0.20/$0.50 per 1M tokens)
+ * - grok-code-fast-1: Agentic coding ($0.20/$1.50 per 1M tokens)
+ *
+ * Pricing: https://x.ai/api#pricing
+ */
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.LLMxai = void 0;
+exports.LLMxai = {
+    // Vision capable models
+    "VISION": {
+        model: "grok-4",
+        temperature: 0.2,
+        stream: false
+    },
+    "VISION-fast": {
+        model: "grok-4-fast-reasoning",
+        temperature: 0.2,
+        stream: false
+    },
+    // Embedding models - xAI ne supporte pas les embeddings
+    // Si utilisé, une erreur sera levée par modelConfig
+    // "EMBEDDING-small": undefined,
+    // "EMBEDDING-large": undefined,
+    // Audio models - xAI ne supporte pas Whisper
+    // "WHISPER": undefined,
+    // "TRANSCRIBE": undefined,
+    // Chat/Responses models
+    "LOW-fast": {
+        temperature: 0.2,
+        model: "grok-code-fast-1",
+        stream: true
+    },
+    "LOW": {
+        temperature: 0.2,
+        model: "grok-4-fast-non-reasoning",
+        stream: true
+    },
+    "LOW-medium": {
+        temperature: 0.2,
+        model: "grok-4-fast-non-reasoning",
+        stream: true
+    },
+    "MEDIUM-fast": {
+        temperature: 0.2,
+        model: "grok-4-1-fast-reasoning",
+        stream: true
+    },
+    "MEDIUM": {
+        temperature: 0.2,
+        model: "grok-4-fast-reasoning",
+        stream: true
+    },
+    "HIGH-fast": {
+        model: "grok-4-1-fast-reasoning",
+        temperature: 0.2,
+        stream: true
+    },
+    "HIGH": {
+        model: "grok-4",
+        temperature: 0.2,
+        stream: true
+    },
+    "HIGH-medium": {
+        model: "grok-4",
+        temperature: 0.2,
+        stream: true
+    },
+    "SEARCH-fast": {
+        temperature: 0.2,
+        model: "grok-code-fast-1",
+        tools: [{ type: "web_search" }],
+    },
+    "SEARCH": {
+        temperature: 0.2,
+        model: "grok-4-fast-reasoning",
+        tools: [{ type: "web_search" }],
+    },
+};

package/dist/src/pricing.llm.d.ts

@@ -1,17 +1,5 @@
-import { CompletionUsage } from "openai/resources";
-import { Usage } from "./types";
-type ModelPricing = {
-    input: number;
-    cachedInput?: number;
-    output: number;
-};
-export declare const modelPricing: Record<string, ModelPricing>;
-export declare function calculateCost(model: string, usage?: CompletionUsage): number;
-export declare function accumulateCost(currentUsage: Usage, model: string, usage?: CompletionUsage): number;
-export declare function LLM(openai: any): any;
-export declare const LLMxai: any;
 /**
- * Get model mapping for OpenAI
+ * @deprecated Ce fichier est conservé pour rétrocompatibilité.
+ * Utiliser les imports depuis './llm/pricing' directement.
  */
-export declare const LLMopenai: any;
-export {};
+export { modelPricing, calculateCost, accumulateCost, LLM, LLMxai, LLMopenai, } from './llm/pricing';

package/dist/src/pricing.llm.js

@@ -1,234 +1,14 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.LLMopenai = exports.LLMxai = exports.modelPricing = void 0;
-exports.calculateCost = calculateCost;
-exports.accumulateCost = accumulateCost;
-exports.LLM = LLM;
-//
-// get/update pricing from openai
-// - https://platform.openai.com/docs/pricing#latest-models
-exports.modelPricing = {
-    "gpt-4.5-preview": { input: 0.000075, cachedInput: 0.0000325, output: 0.000125 },
-    "gpt-4.1": { input: 0.000002, cachedInput: 0.0000005, output: 0.000008 },
-    "gpt-4.1-mini": { input: 0.0000004, cachedInput: 0.0000001, output: 0.0000016 },
-    "gpt-4.1-nano": { input: 0.0000001, cachedInput: 0.000000025, output: 0.0000004 },
-    "gpt-4o": { input: 0.0000025, cachedInput: 0.00000125, output: 0.00001 },
-    "gpt-4o-audio-preview": { input: 0.0000025, output: 0.00001 },
-    "gpt-4o-realtime-preview": { input: 0.000005, cachedInput: 0.0000025, output: 0.00002 },
-    "gpt-4o-search-preview": { input: 0.000005, cachedInput: 0.0000025, output: 0.00002 },
-    "gpt-4o-mini": { input: 0.00000015, cachedInput: 0.000000075, output: 0.0000006 },
-    "gpt-4o-mini-audio-preview": { input: 0.00000015, output: 0.0000006 },
-    "gpt-4o-mini-realtime-preview": { input: 0.0000006, cachedInput: 0.0000003, output: 0.0000024 },
-    "gpt-4o-mini-search-preview": { input: 0.0000015, cachedInput: 0.00000075, output: 0.000006 },
-    // GPT-5 family
-    "gpt-5": { input: 0.00000125, output: 0.00001 },
-    "gpt-5-mini": { input: 0.00000025, output: 0.000002 },
-    "gpt-5-nano": { input: 0.00000005, output: 0.0000004 },
-    "o1": { input: 0.000015, cachedInput: 0.0000075, output: 0.00006 },
-    "o4-mini": { input: 0.0000011, cachedInput: 0.00000055, output: 0.0000044 },
-    "o3-mini": { input: 0.0000011, cachedInput: 0.00000055, output: 0.0000044 },
-    "o1-mini": { input: 0.0000011, cachedInput: 0.00000055, output: 0.0000044 },
-    // Grok 4 family - Official pricing from xAI docs (per million tokens)
-    "grok-4-fast-reasoning": { input: 0.0002, cachedInput: 0.0001, output: 0.0005 },
-    "grok-4-fast-non-reasoning": { input: 0.0002, cachedInput: 0.0001, output: 0.0005 },
-    "grok-4-0709": { input: 0.003, cachedInput: 0.0015, output: 0.015 },
-    "grok-code-fast-1": { input: 0.0002, cachedInput: 0.0001, output: 0.0015 },
-};
-function calculateCost(model, usage) {
-    if (!usage) {
-        return 0;
-    }
-    if (!exports.modelPricing[model]) {
-        throw new Error("Unknown model");
-    }
-    const pricing = exports.modelPricing[model];
-    const cost = usage.prompt_tokens * pricing.input +
-        // usage.completion_tokens * (pricing.cachedInput || 0) +
-        usage.completion_tokens * pricing.output;
-    return cost;
-}
-function accumulateCost(currentUsage, model, usage) {
-    if (!usage) {
-        return 0;
-    }
-    currentUsage.prompt += usage.prompt_tokens || 0;
-    currentUsage.completion += usage.completion_tokens || 0;
-    currentUsage.total += usage.total_tokens || 0;
-    const cost = calculateCost(model, usage);
-    currentUsage.cost += cost;
-    return currentUsage.cost;
-}
-//
-// depending on the API source, return the correct mapping between ALIAS and destination models and options.
-// - LOW-fast: openai => gpt-5-nano, xAI => grok-4-nano
-// - etc.
-function LLM(openai) {
-    //
-    // Detect provider based on baseURL
-    const mapping = openai?.baseURL ? LLMmapping[openai?.baseURL] : LLMmapping["default"];
-    return mapping;
-}
-exports.LLMxai = {
-    "LOW-fast": {
-        temperature: 0.2,
-        model: "grok-code-fast-1", // Fast code model, good for low-complexity tasks
-        stream: true
-    },
-    "LOW": {
-        temperature: 0.2,
-        model: "grok-4-fast-non-reasoning", // Fast non-reasoning model
-        stream: true
-    },
-    "MEDIUM-fast": {
-        temperature: 0.2,
-        model: "grok-4-fast-reasoning", // Fast reasoning model
-        stream: true
-    },
-    "MEDIUM": {
-        temperature: 0.2,
-        model: "grok-4-fast-reasoning", // Fast reasoning model
-        stream: true
-    },
-    "HIGH-fast": {
-        model: "grok-4-fast-reasoning",
-        temperature: 0.2,
-        stream: true
-    },
-    "HIGH-low": {
-        model: "grok-4-0709", // Most capable Grok-4 model
-        temperature: 0.2,
-        stream: true
-    },
-    "HIGH-medium": {
-        model: "grok-4-0709", // Most capable Grok-4 model
-        temperature: 0.2,
-        stream: true
-    },
-    "SEARCH-fast": {
-        temperature: 0.2,
-        model: "grok-code-fast-1",
-        web_search_options: {
-            user_location: {
-                type: "approximate",
-                approximate: {
-                    country: "CH",
-                    city: "Geneva",
-                    region: "Geneva",
-                },
-            },
-        },
-    },
-    "SEARCH": {
-        temperature: 0.2,
-        model: "grok-4-fast-non-reasoning",
-        web_search_options: {
-            user_location: {
-                type: "approximate",
-                approximate: {
-                    country: "CH",
-                    city: "Geneva",
-                    region: "Geneva",
-                },
-            },
-        },
-    },
-};
+exports.LLMopenai = exports.LLMxai = exports.LLM = exports.accumulateCost = exports.calculateCost = exports.modelPricing = void 0;
 /**
- * Get model mapping for OpenAI
+ * @deprecated Ce fichier est conservé pour rétrocompatibilité.
+ * Utiliser les imports depuis './llm/pricing' directement.
  */
-exports.LLMopenai = {
-    "LOW-fast": {
-        temperature: 1,
-        frequency_penalty: 0.0,
-        presence_penalty: 0.0,
-        model: "gpt-5-nano",
-        reasoning_effort: "minimal",
-        verbosity: "low",
-        stream: true
-    },
-    "LOW": {
-        temperature: 1,
-        frequency_penalty: 0.0,
-        presence_penalty: 0.0,
-        model: "gpt-5-nano",
-        reasoning_effort: "medium",
-        verbosity: "low",
-        stream: true
-    },
-    "MEDIUM-fast": {
-        temperature: 1,
-        frequency_penalty: 0.0,
-        presence_penalty: 0.0,
-        model: "gpt-5-mini",
-        reasoning_effort: "minimal",
-        verbosity: "low",
-        stream: true
-    },
-    "LOW-4.1": {
-        temperature: .2,
-        frequency_penalty: 0.0,
-        presence_penalty: 0.0,
-        model: "gpt-4.1-nano",
-        stream: true
-    },
-    "MEDIUM-4.1-mini": {
-        temperature: .2,
-        frequency_penalty: 0.0,
-        presence_penalty: 0.0,
-        model: "gpt-4.1-mini",
-        stream: true
-    },
-    "MEDIUM-4.1": {
-        temperature: .2,
-        frequency_penalty: 0.0,
-        presence_penalty: 0.0,
-        model: "gpt-4.1",
-        stream: true
-    },
-    "MEDIUM": {
-        temperature: 1,
-        frequency_penalty: 0.0,
-        presence_penalty: 0.0,
-        model: "gpt-5-mini",
-        reasoning_effort: "low",
-        verbosity: "low",
-        stream: true
-    },
-    "HIGH-fast": {
-        model: "gpt-5",
-        reasoning_effort: "minimal",
-        verbosity: "low",
-        temperature: 1,
-        stream: true
-    },
-    "HIGH-low": {
-        model: "gpt-5",
-        reasoning_effort: "low",
-        verbosity: "low",
-        stream: true
-    },
-    "HIGH-medium": {
-        model: "gpt-5",
-        reasoning_effort: "medium",
-        verbosity: "low",
-        stream: true
-    },
-    "SEARCH": {
-        model: "gpt-4o-mini-search-preview",
-        web_search_options: {
-            user_location: {
-                type: "approximate",
-                approximate: {
-                    country: "CH",
-                    city: "Geneva",
-                    region: "Geneva",
-                },
-            },
-        },
-    },
-};
-const LLMmapping = {
-    "https://api.x.ai/v1": exports.LLMxai,
-    "https://api.openai.com/v1": exports.LLMopenai,
-    "default": exports.LLMopenai
-};
+var pricing_1 = require("./llm/pricing");
+Object.defineProperty(exports, "modelPricing", { enumerable: true, get: function () { return pricing_1.modelPricing; } });
+Object.defineProperty(exports, "calculateCost", { enumerable: true, get: function () { return pricing_1.calculateCost; } });
+Object.defineProperty(exports, "accumulateCost", { enumerable: true, get: function () { return pricing_1.accumulateCost; } });
+Object.defineProperty(exports, "LLM", { enumerable: true, get: function () { return pricing_1.LLM; } });
+Object.defineProperty(exports, "LLMxai", { enumerable: true, get: function () { return pricing_1.LLMxai; } });
+Object.defineProperty(exports, "LLMopenai", { enumerable: true, get: function () { return pricing_1.LLMopenai; } });

package/dist/src/prompts.d.ts

@@ -11,5 +11,4 @@ export interface UsecaseExtractionOptions {
     examples?: string;
     extended?: string;
 }
-export declare const usecaseExtractionPrompt_OLD: (file: string) => string;
 export declare const usecaseExtractionPrompt: (file: string, options?: UsecaseExtractionOptions) => string;