@decido/kernel-bridge 1.0.0 → 4.0.2

package/src/ai/services/TokenWallet.ts

@@ -1,237 +0,0 @@
-/**
- * TokenWallet — Usage tracking & cost estimation for LLM providers
- *
- * Records every inference call and calculates estimated costs
- * using a per-model pricing table. Persists via localStorage
- * (with Tauri fs fallback).
- */
-
-// ─── Pricing Table (USD per 1M tokens) ─────────────────────
-
-interface ModelPricing {
-  input: number;   // per 1M input tokens
-  output: number;  // per 1M output tokens
-}
-
-const PRICING: Record<string, ModelPricing> = {
-  // Ollama / local — free
-  'qwen2:latest': { input: 0, output: 0 },
-  'llama3:latest': { input: 0, output: 0 },
-  'mistral:latest': { input: 0, output: 0 },
-  'codellama:latest': { input: 0, output: 0 },
-
-  // Gemini
-  'gemini-flash-lite-latest': { input: 0.10, output: 0.40 },
-  'gemini-flash-lite-latest-lite': { input: 0.075, output: 0.30 },
-  'gemini-1.5-pro': { input: 1.25, output: 5.00 },
-  'gemini-flash-latest': { input: 0.075, output: 0.30 },
-
-  // Anthropic
-  'claude-sonnet-4-20250514': { input: 3.00, output: 15.00 },
-  'claude-3-5-sonnet-20241022': { input: 3.00, output: 15.00 },
-  'claude-3-5-haiku-20241022': { input: 0.80, output: 4.00 },
-  'claude-3-opus-20240229': { input: 15.00, output: 75.00 },
-
-  // OpenAI
-  'gpt-4o': { input: 2.50, output: 10.00 },
-  'gpt-4o-mini': { input: 0.15, output: 0.60 },
-  'gpt-4-turbo': { input: 10.00, output: 30.00 },
-  'gpt-4': { input: 30.00, output: 60.00 },
-  'gpt-3.5-turbo': { input: 0.50, output: 1.50 },
-  'o3-mini': { input: 1.10, output: 4.40 },
-};
-
-// Default pricing for unknown models (assume cloud provider)
-const DEFAULT_PRICING: ModelPricing = { input: 1.00, output: 3.00 };
-
-// ─── Types ──────────────────────────────────────────────────
-
-export interface TokenUsageEntry {
-  provider: string;
-  model: string;
-  inputTokens: number;
-  outputTokens: number;
-  totalTokens: number;
-  estimatedCostUsd: number;
-  timestamp: number;
-}
-
-export interface ProviderSummary {
-  tokens: number;
-  cost: number;
-  calls: number;
-}
-
-export interface WalletSummary {
-  totalTokens: number;
-  totalCostUsd: number;
-  totalCalls: number;
-  byProvider: Record<string, ProviderSummary>;
-  sessionStart: number;
-}
-
-type WalletListener = (summary: WalletSummary) => void;
-
-// ─── Storage Key ────────────────────────────────────────────
-
-const STORAGE_KEY = 'decido-token-wallet';
-const MAX_HISTORY = 500; // keep last 500 entries
-
-// ─── TokenWallet Class ──────────────────────────────────────
-
-class TokenWalletImpl {
-  private history: TokenUsageEntry[] = [];
-  private sessionStart = Date.now();
-  private listeners = new Set<WalletListener>();
-
-  constructor() {
-    this._loadFromStorage();
-  }
-
-  // ── Record Usage ────────────────────────────────────────
-
-  record(entry: {
-    provider: string;
-    model: string;
-    tokensUsed?: number;
-    inputTokens?: number;
-    outputTokens?: number;
-    latencyMs?: number;
-  }): void {
-    // Estimate input/output split if not provided
-    const totalTokens = entry.tokensUsed ?? (entry.inputTokens ?? 0) + (entry.outputTokens ?? 0);
-    const inputTokens = entry.inputTokens ?? Math.round(totalTokens * 0.4);
-    const outputTokens = entry.outputTokens ?? (totalTokens - inputTokens);
-
-    const pricing = this._getPricing(entry.model, entry.provider);
-    const estimatedCostUsd =
-      (inputTokens / 1_000_000) * pricing.input +
-      (outputTokens / 1_000_000) * pricing.output;
-
-    const usageEntry: TokenUsageEntry = {
-      provider: entry.provider,
-      model: entry.model,
-      inputTokens,
-      outputTokens,
-      totalTokens,
-      estimatedCostUsd,
-      timestamp: Date.now(),
-    };
-
-    this.history.push(usageEntry);
-
-    // Trim history if too large
-    if (this.history.length > MAX_HISTORY) {
-      this.history = this.history.slice(-MAX_HISTORY);
-    }
-
-    this._saveToStorage();
-    this._emit();
-
-    console.log(
-      `💰 [Wallet] ${entry.provider}/${entry.model}: ${totalTokens} tokens ≈ $${estimatedCostUsd.toFixed(6)}`
-    );
-  }
-
-  // ── Summary ─────────────────────────────────────────────
-
-  getSummary(): WalletSummary {
-    const byProvider: Record<string, ProviderSummary> = {};
-    let totalTokens = 0;
-    let totalCostUsd = 0;
-
-    for (const entry of this.history) {
-      totalTokens += entry.totalTokens;
-      totalCostUsd += entry.estimatedCostUsd;
-
-      if (!byProvider[entry.provider]) {
-        byProvider[entry.provider] = { tokens: 0, cost: 0, calls: 0 };
-      }
-      byProvider[entry.provider].tokens += entry.totalTokens;
-      byProvider[entry.provider].cost += entry.estimatedCostUsd;
-      byProvider[entry.provider].calls += 1;
-    }
-
-    return {
-      totalTokens,
-      totalCostUsd,
-      totalCalls: this.history.length,
-      byProvider,
-      sessionStart: this.sessionStart,
-    };
-  }
-
-  getHistory(): TokenUsageEntry[] {
-    return [...this.history];
-  }
-
-  getRecentHistory(count: number): TokenUsageEntry[] {
-    return this.history.slice(-count);
-  }
-
-  // ── Lifecycle ───────────────────────────────────────────
-
-  clearHistory(): void {
-    this.history = [];
-    this.sessionStart = Date.now();
-    this._saveToStorage();
-    this._emit();
-  }
-
-  // ── Subscriptions ───────────────────────────────────────
-
-  subscribe(listener: WalletListener): () => void {
-    this.listeners.add(listener);
-    return () => { this.listeners.delete(listener); };
-  }
-
-  // ── Pricing Lookup ──────────────────────────────────────
-
-  private _getPricing(model: string, provider: string): ModelPricing {
-    // Exact match
-    if (PRICING[model]) return PRICING[model];
-
-    // Local providers are always free
-    if (provider === 'ollama' || provider === 'mlx') {
-      return { input: 0, output: 0 };
-    }
-
-    return DEFAULT_PRICING;
-  }
-
-  // ── Persistence ─────────────────────────────────────────
-
-  private _saveToStorage(): void {
-    try {
-      const data = JSON.stringify({
-        history: this.history,
-        sessionStart: this.sessionStart,
-      });
-      localStorage.setItem(STORAGE_KEY, data);
-    } catch { /* storage full or unavailable */ }
-  }
-
-  private _loadFromStorage(): void {
-    try {
-      const raw = localStorage.getItem(STORAGE_KEY);
-      if (raw) {
-        const data = JSON.parse(raw);
-        this.history = data.history || [];
-        this.sessionStart = data.sessionStart || Date.now();
-      }
-    } catch { /* corrupted data, start fresh */ }
-  }
-
-  // ── Emit ────────────────────────────────────────────────
-
-  private _emit(): void {
-    const summary = this.getSummary();
-    for (const listener of this.listeners) {
-      try { listener(summary); } catch { /* ignore listener errors */ }
-    }
-  }
-}
-
-// ─── Singleton Export ───────────────────────────────────────
-
-export const tokenWallet = new TokenWalletImpl();

package/src/ai/services/providers/AnthropicProvider.ts

@@ -1,229 +0,0 @@
-/**
- * AnthropicProvider — Claude LLM provider
- * 
- * Supports Claude 3.5 Sonnet, Claude 3 Opus/Haiku via Anthropic Messages API.
- * Implements both streaming and non-streaming chat.
- */
-
-import type { LLMProvider, ChatMessage, ChatOptions, ChatResult, ProviderStatus } from './LLMProvider';
-
-// ─── Config ─────────────────────────────────────────────────
-
-const ANTHROPIC_API_BASE = 'https://api.anthropic.com/v1';
-const DEFAULT_MODEL = 'claude-sonnet-4-20250514';
-const API_VERSION = '2023-06-01';
-
-const AVAILABLE_MODELS = [
-    'claude-sonnet-4-20250514',
-    'claude-3-5-sonnet-20241022',
-    'claude-3-5-haiku-20241022',
-    'claude-3-opus-20240229',
-];
-
-// ─── Provider Implementation ────────────────────────────────
-
-export class AnthropicProvider implements LLMProvider {
-    readonly id = 'anthropic';
-    readonly name = 'Anthropic Claude';
-    readonly requiresApiKey = true;
-    readonly defaultModel = DEFAULT_MODEL;
-
-    private apiKey: string | null = null;
-
-    setApiKey(key: string): void {
-        this.apiKey = key;
-    }
-
-    async listModels(): Promise<string[]> {
-        return AVAILABLE_MODELS;
-    }
-
-    async checkStatus(): Promise<ProviderStatus> {
-        if (!this.apiKey) return 'unconfigured';
-
-        try {
-            // Quick health check — list models endpoint
-            const res = await fetch(`${ANTHROPIC_API_BASE}/messages`, {
-                method: 'POST',
-                headers: {
-                    'x-api-key': this.apiKey,
-                    'anthropic-version': API_VERSION,
-                    'content-type': 'application/json',
-                },
-                body: JSON.stringify({
-                    model: this.defaultModel,
-                    max_tokens: 1,
-                    messages: [{ role: 'user', content: 'ping' }],
-                }),
-                signal: AbortSignal.timeout(5000),
-            });
-            return res.ok || res.status === 429 ? 'available' : 'error';
-        } catch {
-            return 'unavailable';
-        }
-    }
-
-    async chat(messages: ChatMessage[], options?: ChatOptions): Promise<ChatResult> {
-        if (!this.apiKey) {
-            throw new Error('Anthropic API key not configured. Set it in Settings → Providers.');
-        }
-
-        const model = options?.model || this.defaultModel;
-        const start = Date.now();
-
-        // Separate system from conversation
-        const systemMsg = messages.find(m => m.role === 'system');
-        const conversationMsgs = messages
-            .filter(m => m.role !== 'system')
-            .map(m => ({
-                role: m.role as 'user' | 'assistant',
-                content: m.content,
-            }));
-
-        // Ensure alternating user/assistant messages (Anthropic requirement)
-        const sanitized = this._sanitizeMessages(conversationMsgs);
-
-        const body: Record<string, unknown> = {
-            model,
-            max_tokens: options?.maxTokens ?? 2048,
-            messages: sanitized,
-        };
-
-        if (options?.temperature !== undefined) {
-            body.temperature = options.temperature;
-        }
-
-        if (systemMsg) {
-            body.system = systemMsg.content;
-        }
-
-        const res = await fetch(`${ANTHROPIC_API_BASE}/messages`, {
-            method: 'POST',
-            headers: {
-                'x-api-key': this.apiKey,
-                'anthropic-version': API_VERSION,
-                'content-type': 'application/json',
-            },
-            body: JSON.stringify(body),
-            signal: options?.signal ?? AbortSignal.timeout(60000),
-        });
-
-        if (!res.ok) {
-            const errorBody = await res.text();
-            throw new Error(`Anthropic API error ${res.status}: ${errorBody}`);
-        }
-
-        const data = await res.json();
-        const latencyMs = Date.now() - start;
-
-        // Extract text from content blocks
-        const text = (data.content || [])
-            .filter((block: { type: string }) => block.type === 'text')
-            .map((block: { text: string }) => block.text)
-            .join('') || '(sin respuesta)';
-
-        const tokensUsed =
-            (data.usage?.input_tokens || 0) + (data.usage?.output_tokens || 0);
-
-        return { text, model, latencyMs, tokensUsed };
-    }
-
-    /** Stream chat completion — yields text chunks */
-    async *chatStream(
-        messages: ChatMessage[],
-        options?: ChatOptions,
-    ): AsyncGenerator<string, void, unknown> {
-        if (!this.apiKey) {
-            throw new Error('Anthropic API key not configured.');
-        }
-
-        const model = options?.model || this.defaultModel;
-        const systemMsg = messages.find(m => m.role === 'system');
-        const conversationMsgs = messages
-            .filter(m => m.role !== 'system')
-            .map(m => ({ role: m.role as 'user' | 'assistant', content: m.content }));
-
-        const sanitized = this._sanitizeMessages(conversationMsgs);
-
-        const body: Record<string, unknown> = {
-            model,
-            max_tokens: options?.maxTokens ?? 2048,
-            messages: sanitized,
-            stream: true,
-        };
-
-        if (options?.temperature !== undefined) body.temperature = options.temperature;
-        if (systemMsg) body.system = systemMsg.content;
-
-        const res = await fetch(`${ANTHROPIC_API_BASE}/messages`, {
-            method: 'POST',
-            headers: {
-                'x-api-key': this.apiKey,
-                'anthropic-version': API_VERSION,
-                'content-type': 'application/json',
-            },
-            body: JSON.stringify(body),
-            signal: options?.signal ?? AbortSignal.timeout(120000),
-        });
-
-        if (!res.ok) {
-            const errorBody = await res.text();
-            throw new Error(`Anthropic stream error ${res.status}: ${errorBody}`);
-        }
-
-        const reader = res.body?.getReader();
-        if (!reader) return;
-
-        const decoder = new TextDecoder();
-        let buffer = '';
-
-        while (true) {
-            const { done, value } = await reader.read();
-            if (done) break;
-
-            buffer += decoder.decode(value, { stream: true });
-            const lines = buffer.split('\n');
-            buffer = lines.pop() || '';
-
-            for (const line of lines) {
-                if (!line.startsWith('data: ')) continue;
-                const data = line.slice(6).trim();
-                if (data === '[DONE]') return;
-
-                try {
-                    const event = JSON.parse(data);
-                    if (event.type === 'content_block_delta' && event.delta?.text) {
-                        yield event.delta.text;
-                    }
-                } catch { /* skip malformed lines */ }
-            }
-        }
-    }
-
-    /** Ensure alternating user/assistant (Anthropic requires this) */
-    private _sanitizeMessages(
-        msgs: Array<{ role: string; content: string }>,
-    ): Array<{ role: string; content: string }> {
-        if (msgs.length === 0) return [{ role: 'user', content: '...' }];
-
-        const result: Array<{ role: string; content: string }> = [];
-        let lastRole = '';
-
-        for (const msg of msgs) {
-            if (msg.role === lastRole) {
-                // Merge consecutive same-role messages
-                result[result.length - 1].content += '\n' + msg.content;
-            } else {
-                result.push({ ...msg });
-                lastRole = msg.role;
-            }
-        }
-
-        // Must start with 'user'
-        if (result[0]?.role !== 'user') {
-            result.unshift({ role: 'user', content: '...' });
-        }
-
-        return result;
-    }
-}

package/src/ai/services/providers/GeminiProvider.ts

@@ -1,121 +0,0 @@
-/**
- * GeminiProvider — Google Gemini LLM provider
- * 
- * Uses the Gemini REST API directly (no SDK dependency).
- * Supports gemini-flash-lite-latest, gemini-1.5-pro, etc.
- */
-
-import type { LLMProvider, ChatMessage, ChatOptions, ChatResult, ProviderStatus } from './LLMProvider';
-
-// ─── Config ─────────────────────────────────────────────────
-
-const GEMINI_API_BASE = 'https://generativelanguage.googleapis.com/v1beta';
-const DEFAULT_MODEL = 'gemini-flash-lite-latest';
-
-const AVAILABLE_MODELS = [
-    'gemini-flash-lite-latest',
-    'gemini-flash-lite-latest-lite',
-    'gemini-1.5-pro',
-    'gemini-flash-latest',
-];
-
-// ─── Provider Implementation ────────────────────────────────
-
-export class GeminiProvider implements LLMProvider {
-    readonly id = 'gemini';
-    readonly name = 'Google Gemini';
-    readonly requiresApiKey = true;
-    readonly defaultModel = DEFAULT_MODEL;
-
-    private apiKey: string | null = null;
-
-    setApiKey(key: string): void {
-        this.apiKey = key;
-    }
-
-    async listModels(): Promise<string[]> {
-        return AVAILABLE_MODELS;
-    }
-
-    async checkStatus(): Promise<ProviderStatus> {
-        if (!this.apiKey) return 'unconfigured';
-
-        try {
-            // Quick validation — list models endpoint
-            const res = await fetch(
-                `${GEMINI_API_BASE}/models?key=${this.apiKey}`,
-                { signal: AbortSignal.timeout(5000) }
-            );
-            return res.ok ? 'available' : 'error';
-        } catch {
-            return 'unavailable';
-        }
-    }
-
-    async chat(messages: ChatMessage[], options?: ChatOptions): Promise<ChatResult> {
-        if (!this.apiKey) {
-            throw new Error('Gemini API key not configured. Set it in Settings → Providers.');
-        }
-
-        const model = options?.model || this.defaultModel;
-        const start = Date.now();
-
-        // Separate system instruction from conversation
-        const systemMsg = messages.find(m => m.role === 'system');
-        const conversationMsgs = messages.filter(m => m.role !== 'system');
-
-        // Convert to Gemini format
-        const contents = conversationMsgs.map(m => ({
-            role: m.role === 'assistant' ? 'model' : 'user',
-            parts: [{ text: m.content }],
-        }));
-
-        const body: Record<string, unknown> = {
-            contents,
-            generationConfig: {
-                temperature: options?.temperature ?? 0.7,
-                maxOutputTokens: options?.maxTokens ?? 2048,
-            },
-        };
-
-        if (systemMsg) {
-            body.systemInstruction = {
-                parts: [{ text: systemMsg.content }],
-            };
-        }
-
-        const url = `${GEMINI_API_BASE}/models/${model}:generateContent?key=${this.apiKey}`;
-
-        const res = await fetch(url, {
-            method: 'POST',
-            headers: { 'Content-Type': 'application/json' },
-            body: JSON.stringify(body),
-            signal: options?.signal ?? AbortSignal.timeout(60000),
-        });
-
-        if (!res.ok) {
-            const errorBody = await res.text();
-            throw new Error(`Gemini API error ${res.status}: ${errorBody}`);
-        }
-
-        const data = await res.json();
-        const latencyMs = Date.now() - start;
-
-        // Extract text from response
-        const candidate = data.candidates?.[0];
-        const text = candidate?.content?.parts
-            ?.map((p: { text?: string }) => p.text || '')
-            .join('') || '(sin respuesta)';
-
-        const tokensUsed =
-            (data.usageMetadata?.promptTokenCount || 0) +
-            (data.usageMetadata?.candidatesTokenCount || 0);
-
-        return {
-            text,
-            model,
-            latencyMs,
-            tokensUsed,
-        };
-    }
-}

package/src/ai/services/providers/LLMProvider.ts

@@ -1,72 +0,0 @@
-/**
- * LLMProvider — Universal interface for LLM backends
- * 
- * All providers (Ollama, Gemini, OpenAI, etc.) implement this
- * interface so the InferenceRouter can swap them transparently.
- */
-
-// ─── Types ──────────────────────────────────────────────────
-
-export interface ChatMessage {
-    role: 'system' | 'user' | 'assistant';
-    content: string;
-}
-
-export interface ChatOptions {
-    model?: string;
-    temperature?: number;
-    maxTokens?: number;
-    /** AbortSignal for cancellation */
-    signal?: AbortSignal;
-}
-
-export interface ChatResult {
-    text: string;
-    model: string;
-    latencyMs: number;
-    tokensUsed?: number;
-}
-
-export type ProviderStatus = 'available' | 'unavailable' | 'error' | 'unconfigured';
-
-// ─── Provider Interface ─────────────────────────────────────
-
-export interface LLMProvider {
-    /** Unique identifier */
-    readonly id: string;
-    /** Display name */
-    readonly name: string;
-    /** Whether this provider needs an API key */
-    readonly requiresApiKey: boolean;
-    /** Default model for this provider */
-    readonly defaultModel: string;
-    /** Available models */
-    listModels(): Promise<string[]>;
-    /** Health check */
-    checkStatus(): Promise<ProviderStatus>;
-    /** Send a chat completion request */
-    chat(messages: ChatMessage[], options?: ChatOptions): Promise<ChatResult>;
-    /** Set API key if required */
-    setApiKey?(key: string): void;
-}
-
-// ─── Provider Registry ──────────────────────────────────────
-
-const providers = new Map<string, LLMProvider>();
-
-export function registerProvider(provider: LLMProvider): void {
-    providers.set(provider.id, provider);
-    console.log(`🔌 [LLM] Provider registered: ${provider.name} (${provider.id})`);
-}
-
-export function getProvider(id: string): LLMProvider | undefined {
-    return providers.get(id);
-}
-
-export function getAllProviders(): LLMProvider[] {
-    return Array.from(providers.values());
-}
-
-export function getAvailableProviderIds(): string[] {
-    return Array.from(providers.keys());
-}