Skip to main content

​
API Standards

To ensure consistency, security, and AI-native readiness, all API development in Tonnex must follow these standards.

​
Architecture

We use NestJS as the core logic engine, Drizzle ORM for database access, and Supabase for Auth and RLS.

​
Service Pattern

All services must use the db.rls pattern to ensure multi-tenancy is enforced. 
@Injectable()
export class MyService {
  async performAction(token: string, data: any) {
    const db = createDrizzleSupabaseClient(token);
    return await db.rls(async (tx) => {
      // Logic goes here
    });
  }
}

​
Agentic Tier API Design

To be truly AI-Native, APIs must act as intelligent interfaces for autonomous agents, facilitating reasoning and self-correction.
  1. Semantic Reasoning: Services should return enriched data including a reasoning field explaining why an action was taken (e.g., “Vehicle chosen via proximity algorithm”).
  2. Actionable Feedback (Self-Correction Hints): Error responses should include a correctionHint. Instead of just 400 Bad Request, provide context like "Invalid pincode for the given city. Please correct or provide a valid pincode for [City Name]."
  3. Static Discovery (The Agent Manual): Use deep Swagger (@ApiOperation) instructions as a manual for AI agents to plan their actions before execution.
  4. Standardized Response Envelope:
    • Success: { success: true, data: T, reasoning?: string }
    • Error: { success: false, error: string, correctionHint?: string }
// Standard Agentic Service Pattern
@Injectable()
export class MyService {
  async performAction(token: string, userId: string, data: any) {
    const db = createDrizzleSupabaseClient(token);
    return await db.rls(async (tx) => {
      // 1. Perform Business Logic
      // 2. Return data + reasoning
      return {
        data: { id: "123" },
        reasoning: "Action performed because of condition X."
      };
    });
  }
}