Merge pull request #11 from fusionlabs-tech/dev

Dev

Author: DesmondSanctity

.gitignore

@@ -81,4 +81,8 @@ Thumbs.db
 # Temporary folders
 tmp/
 temp/
-examples/sdk-local-test/
+examples/sdk-local-test/
+
+# MD
+PRODUCTION_ROADMAP.md
+CLAUDE.md

ROADMAP.md

@@ -0,0 +1,23 @@
+-- CreateEnum
+CREATE TYPE "AiProvider" AS ENUM ('OPENAI', 'ANTHROPIC', 'GOOGLE', 'AZURE_OPENAI', 'CUSTOM');
+
+-- CreateTable
+CREATE TABLE "ai_keys" (
+    "id" TEXT NOT NULL,
+    "userId" TEXT NOT NULL,
+    "provider" "AiProvider" NOT NULL,
+    "alias" TEXT,
+    "apiKeyEnc" TEXT NOT NULL,
+    "defaultModel" TEXT,
+    "endpointBase" TEXT,
+    "orgId" TEXT,
+    "isActive" BOOLEAN NOT NULL DEFAULT true,
+    "lastUsedAt" TIMESTAMP(3),
+    "createdAt" TIMESTAMP(3) NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    "updatedAt" TIMESTAMP(3) NOT NULL,
+
+    CONSTRAINT "ai_keys_pkey" PRIMARY KEY ("id")
+);
+
+-- AddForeignKey
+ALTER TABLE "ai_keys" ADD CONSTRAINT "ai_keys_userId_fkey" FOREIGN KEY ("userId") REFERENCES "users"("id") ON DELETE CASCADE ON UPDATE CASCADE;

backend/prisma/migrations/20250810231426_ai_keys/migration.sql

@@ -34,6 +34,7 @@ model User {
   apiKeys       ApiKey[]
   sessions      UserSession[]
   notifications Notification[]
+  aiKeys        AiKey[]
 
   @@map("users")
 }
@@ -170,6 +171,25 @@ model Notification {
   @@map("notifications")
 }
 
+model AiKey {
+  id           String     @id @default(uuid())
+  userId       String
+  provider     AiProvider
+  alias        String?
+  apiKeyEnc    String
+  defaultModel String?
+  endpointBase String?
+  orgId        String?
+  isActive     Boolean    @default(true)
+  lastUsedAt   DateTime?
+  createdAt    DateTime   @default(now())
+  updatedAt    DateTime   @updatedAt
+
+  user User @relation(fields: [userId], references: [id], onDelete: Cascade)
+
+  @@map("ai_keys")
+}
+
 enum Role {
   ADMIN
   USER
@@ -193,3 +213,11 @@ enum OAuthProvider {
   GOOGLE
   GITHUB
 }
+
+enum AiProvider {
+  OPENAI
+  ANTHROPIC
+  GOOGLE
+  AZURE_OPENAI
+  CUSTOM
+}

backend/prisma/schema.prisma

@@ -0,0 +1,259 @@
+import { AiProvider } from '@prisma/client';
+
+export type ProviderConfig = {
+ provider: AiProvider;
+ apiKey: string;
+ endpointBase?: string;
+ defaultModel?: string;
+ orgId?: string;
+};
+
+export interface AIProviderAdapter {
+ chatJSON(
+  prompt: string,
+  opts?: { model?: string; system?: string }
+ ): Promise<any>;
+}
+
+const DEFAULTS = {
+ OPENAI: { base: 'https://api.openai.com/v1', model: 'gpt-4o-mini' },
+ ANTHROPIC: {
+  base: 'https://api.anthropic.com',
+  path: '/v1/messages',
+  model: 'claude-3-5-sonnet-latest',
+  version: process.env.ANTHROPIC_API_VERSION || '2023-06-01',
+ },
+ GOOGLE: {
+  base: 'https://generativelanguage.googleapis.com',
+  model: 'gemini-1.5-flash',
+ },
+ AZURE_OPENAI: {
+  // endpointBase should be like: https://your-resource.openai.azure.com
+  apiVersion: process.env.AZURE_OPENAI_API_VERSION || '2024-02-15-preview',
+ },
+};
+
+async function requestWithTimeout(
+ url: string,
+ init: RequestInit,
+ timeoutMs = 12_000
+): Promise<Response> {
+ const controller = new AbortController();
+ const id = setTimeout(() => controller.abort(), timeoutMs);
+ try {
+  const res = await fetch(url, { ...init, signal: controller.signal });
+  return res;
+ } finally {
+  clearTimeout(id);
+ }
+}
+
+async function withRetry<T>(fn: () => Promise<T>, attempts = 2): Promise<T> {
+ let lastErr: any;
+ for (let i = 0; i < attempts; i++) {
+  try {
+   return await fn();
+  } catch (e) {
+   lastErr = e;
+   await new Promise((r) => setTimeout(r, (i + 1) * 300));
+  }
+ }
+ throw lastErr;
+}
+
+export function createAIProviderAdapter(
+ cfg: ProviderConfig
+): AIProviderAdapter {
+ if (cfg.provider === 'OPENAI') {
+  const base = cfg.endpointBase || DEFAULTS.OPENAI.base;
+  const model = cfg.defaultModel || DEFAULTS.OPENAI.model;
+  return {
+   async chatJSON(prompt, opts) {
+    const sys =
+     opts?.system ||
+     'You are a helpful analysis assistant. Output strictly valid JSON only.';
+    const body = {
+     model: opts?.model || model,
+     messages: [
+      { role: 'system', content: sys },
+      { role: 'user', content: prompt },
+     ],
+     temperature: 0.2,
+     response_format: { type: 'json_object' },
+    };
+    const res = await withRetry(() =>
+     requestWithTimeout(`${base}/chat/completions`, {
+      method: 'POST',
+      headers: {
+       'Content-Type': 'application/json',
+       Authorization: `Bearer ${cfg.apiKey}`,
+       ...(cfg.orgId ? { 'OpenAI-Organization': cfg.orgId } : {}),
+      },
+      body: JSON.stringify(body),
+     })
+    );
+    if (!res.ok) throw new Error(`OpenAI error ${res.status}`);
+    const data: any = await res.json();
+    const choices: any[] | undefined = Array.isArray((data as any)?.choices)
+     ? (data as any).choices
+     : undefined;
+    const text =
+     choices && choices[0] && choices[0].message?.content
+      ? String(choices[0].message.content)
+      : '{}';
+    try {
+     return JSON.parse(text);
+    } catch {
+     return { raw: text };
+    }
+   },
+  };
+ }
+
+ if (cfg.provider === 'ANTHROPIC') {
+  const base = cfg.endpointBase || DEFAULTS.ANTHROPIC.base;
+  const path = DEFAULTS.ANTHROPIC.path;
+  const model = cfg.defaultModel || DEFAULTS.ANTHROPIC.model;
+  const version = DEFAULTS.ANTHROPIC.version;
+  return {
+   async chatJSON(prompt, _opts) {
+    const body = {
+     model,
+     max_tokens: 1024,
+     messages: [{ role: 'user', content: prompt }],
+    } as any;
+    const res = await withRetry(() =>
+     requestWithTimeout(`${base}${path}`, {
+      method: 'POST',
+      headers: {
+       'Content-Type': 'application/json',
+       'x-api-key': cfg.apiKey,
+       'anthropic-version': version,
+      },
+      body: JSON.stringify(body),
+     })
+    );
+    if (!res.ok) throw new Error(`Anthropic error ${res.status}`);
+    const data: any = await res.json();
+    const text: string =
+     data?.content?.[0]?.text || data?.content?.[0]?.content?.[0]?.text || '{}';
+    try {
+     return JSON.parse(text);
+    } catch {
+     return { raw: text };
+    }
+   },
+  };
+ }
+
+ if (cfg.provider === 'GOOGLE') {
+  const base = cfg.endpointBase || DEFAULTS.GOOGLE.base;
+  const model = cfg.defaultModel || DEFAULTS.GOOGLE.model;
+  return {
+   async chatJSON(prompt, _opts) {
+    // Google uses API key in query string
+    const url = `${base}/v1beta/models/${encodeURIComponent(model)}:generateContent?key=${encodeURIComponent(cfg.apiKey)}`;
+    const body = {
+     contents: [{ role: 'user', parts: [{ text: prompt }] }],
+     generationConfig: { temperature: 0.2 },
+    } as any;
+    const res = await withRetry(() =>
+     requestWithTimeout(url, {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify(body),
+     })
+    );
+    if (!res.ok) throw new Error(`Google error ${res.status}`);
+    const data: any = await res.json();
+    const text: string =
+     data?.candidates?.[0]?.content?.parts?.[0]?.text || '{}';
+    try {
+     return JSON.parse(text);
+    } catch {
+     return { raw: text };
+    }
+   },
+  };
+ }
+
+ if (cfg.provider === 'AZURE_OPENAI') {
+  const base = cfg.endpointBase?.replace(/\/$/, '') || '';
+  const deployment = cfg.defaultModel; // here defaultModel should be deployment name
+  const apiVersion = DEFAULTS.AZURE_OPENAI.apiVersion;
+  if (!base || !deployment) {
+   return {
+    async chatJSON() {
+     throw new Error(
+      'Azure OpenAI requires endpointBase and defaultModel (deployment name)'
+     );
+    },
+   };
+  }
+  return {
+   async chatJSON(prompt, opts) {
+    const sys =
+     opts?.system ||
+     'You are a helpful analysis assistant. Output strictly valid JSON only.';
+    const body = {
+     messages: [
+      { role: 'system', content: sys },
+      { role: 'user', content: prompt },
+     ],
+     temperature: 0.2,
+     response_format: { type: 'json_object' },
+    } as any;
+    const url = `${base}/openai/deployments/${encodeURIComponent(deployment)}/chat/completions?api-version=${encodeURIComponent(apiVersion)}`;
+    const res = await withRetry(() =>
+     requestWithTimeout(url, {
+      method: 'POST',
+      headers: {
+       'Content-Type': 'application/json',
+       'api-key': cfg.apiKey,
+      },
+      body: JSON.stringify(body),
+     })
+    );
+    if (!res.ok) throw new Error(`Azure OpenAI error ${res.status}`);
+    const data: any = await res.json();
+    const text = data?.choices?.[0]?.message?.content || '{}';
+    try {
+     return JSON.parse(text);
+    } catch {
+     return { raw: text };
+    }
+   },
+  };
+ }
+
+ // Custom HTTP adapter: POST to endpointBase with { prompt }, expect JSON response
+ if (cfg.provider === 'CUSTOM') {
+  const base = cfg.endpointBase || '';
+  return {
+   async chatJSON(prompt) {
+    if (!base) throw new Error('Custom provider requires endpointBase');
+    const res = await withRetry(() =>
+     requestWithTimeout(base, {
+      method: 'POST',
+      headers: {
+       'Content-Type': 'application/json',
+       Authorization: `Bearer ${cfg.apiKey}`,
+      },
+      body: JSON.stringify({ prompt, mode: 'json' }),
+     })
+    );
+    if (!res.ok) throw new Error(`Custom provider error ${res.status}`);
+    const data: any = await res.json();
+    // Assume provider already returns JSON object
+    return data;
+   },
+  };
+ }
+
+ // Default unsupported provider
+ return {
+  async chatJSON() {
+   throw new Error('Unsupported provider');
+  },
+ };
+}