Inspekt Documentation

Install the package, initialise an Inspekt instance, call connect(), and register the adapter for your framework. That's the entire setup.

1. Install

npm install @inspekt/js

2. Initialise

Create an instance and call connect(). The connect() call is compulsory, it establishes the WebSocket bridge that enables real-time log streaming and AI analysis.

import { Inspekt } from '@inspekt/js';

const inspekt = new Inspekt({
    apiKey: "ins_live_...",   // 57-character unique key
    terminalOutput: true,     // Renders AI diagnosis cards in the console
    analysisMode: 'errors',   // 'errors' | 'always' | 'never'
    redactKeys: ["password"]  // Custom keys to scrub from headers/body
});

// COMPULSORY: Establishes the WebSocket bridge
inspekt.connect();

3. Get your API key

Generate a key from the API Keys section of your dashboard and pass it as apiKey option. Keys follow the format ins_live_... and are 57 characters long.

Inspekt ships a dedicated adapter for each supported framework. Every adapter uses the framework's native hooks to capture the full request-response cycle with zero-latency overhead.

Express

The Express adapter hijacks res.json and listens for the finish event to calculate response times and capture payloads. Mount it early in your middleware stack so it captures all downstream errors.

import { Inspekt } from '@inspekt/js';
import { inspektExpress } from '@inspekt/js/express';
import express from 'express';

const inspekt = new Inspekt({ apiKey: "ins_live_..." });
inspekt.connect();

const app = express();

// Place early in the middleware stack
app.use(inspektExpress(inspekt));

Fastify

Register Inspekt as a plugin to monitor traffic with minimal overhead using Fastify's native onRequest, preSerialization, and onResponse hooks.

import { Inspekt } from '@inspekt/js';
import { inspektFastify } from '@inspekt/js/fastify';
import Fastify from 'fastify';

const inspekt = new Inspekt({ apiKey: "ins_live_..." });
inspekt.connect();

const fastify = Fastify();
fastify.register(inspektFastify(inspekt));

NestJS

Implemented as a NestInterceptor that uses RxJS tap to monitor successful responses and caught exceptions. Provide the Inspekt instance via your module and apply the interceptor at the controller or global level.

import { Module, Controller, Get, UseInterceptors } from '@nestjs/common';
import { NestFactory } from '@nestjs/core';
import { Inspekt } from '@inspekt/js';
import { InspektInterceptor } from '@inspekt/js/nest';

const inspekt = new Inspekt({
    apiKey: "ins_live_...",
    // 'authorization', 'cookie' and 'set-cookie' are redacted by default.
    redactKeys: ['x-api-secret', 'password', 'ssn']
});
inspekt.connect()

@Module({
    providers: [
        {
            provide: 'INSPEKT',
            useFactory: inspekt,
        },
    ],
})
export class AppModule { }

@Controller()
@UseInterceptors(InspektInterceptor)
export class AppController {}

async function bootstrap() {
    const app = await NestFactory.create(AppModule);
    await app.listen(3000);
    console.log('🚀 Server ready at http://localhost:3000');
}
bootstrap();

Options

All options are passed to the Inspekt constructor.

Analysis Modes

The analysisMode option allows you to balance detailed observability against your monthly analysis quota.

Redaction

Inspekt redacts the authorization header from all payloads by default before anything is transmitted. Extend this with the redactKeys array to scrub any additional header or body keys.

const inspekt = new Inspekt({
    apiKey: "ins_live_...",
    // 'authorization' is always redacted by default.
    // Add any additional sensitive keys here:
    redactKeys: ["x-api-secret", "password", "token", "ssn"]
});

Real-time Streaming

On connect(), the SDK performs a handshake that synchronises your redactKeys and analysisMode configuration with the socket server. From that point, every intercepted request is streamed in real-time to your dashboard with no polling or batch delays.

Terminal Output

When terminalOutput: true is enabled, Inspekt transforms raw socket data into structured, color-coded diagnosis cards. These cards are rendered immediately upon a successful handshake with the AI engine, providing a real-time debugging interface without leaving your development environment.

Each terminal card decomposes the request into six primary focus areas:

• Summary: A high-level overview of the event categorized by its severity level.
• Technical Diagnosis: A deep-dive analysis of the root cause derived from the HTTP exchange and stack traces.
• Issues and Anomalies: Specific problems identified within the request lifecycle or payload structure.
• Security and Headers: Assessment of notable headers, missing requirements, and active security flags.
• Performance and Logic: Identification of bottlenecks, timing issues, or unexpected status code behaviors.
• Actionable Fixes: Concrete, code-level recommendations to resolve the identified error immediately.

Reconnection

Network stability is rarely guaranteed. If the WebSocket connection drops due to a network blip, server restart, or deployment, Inspekt initiates an automated, infinite recovery sequence using an exponential backoff strategy.

Inspekt uses a capped exponential backoff. The delay between retries doubles with every failure but is capped at a maximum of 30 seconds. This allows the SDK to recover instantly from quick 'blips' while preventing excessive network traffic during long-term outages