Prism

Local-first API workbench, Git-native test collections, and high-concurrency load tester.

Designed by Atlas Research Labs as a zero-bloat, private alternative to Postman and k6.

Architecture Principles

• Local-First & Git-Native: Test suites and collections are flat, version-controlled JSON/YAML files stored directly in your repository. No mandatory accounts, no cloud synchronization lock-in.
• Zero-Overhead Concurrency Engine: Native HTTP/HTTPS connection pooling with socket reuse capable of generating high-throughput stress tests without heavy JVM or Electron resource footprints.
• Microsecond Phase Timings: Detailed telemetry on every request: DNS resolution, TCP handshake, TLS negotiation, Time to First Byte (TTFB), and total response lifecycle.
• Embedded OLED Studio: Pure OLED black local Web UI (http://127.0.0.1:7070) with real-time SSE benchmark streaming, request builder, and assertion reports.
• Zero External Dependencies: Implemented entirely with Node.js standard runtime libraries (http, https, perf_hooks, events). Package size is under 100 KB with idle memory usage under 15 MB.

Installation

Global CLI installation via NPM:

npm install -g prismbench

Or run directly without installation via npx:

npx prismbench --help

Both binary commands prism and prismbench are exposed.

CLI Reference

1. Single Request Execution

Execute an immediate HTTP request with timing analysis:

# Quick request
prism https://api.github.com

# Detailed request with custom method and headers
prism request https://httpbin.org/post -m POST -H "Content-Type: application/json" -b '{"service":"prism"}'

Output includes HTTP status, transfer size, full response headers, formatted body, and a breakdown of DNS, TCP, TLS, TTFB, and total latency.

2. Git-Native Test Collections

Run an automated test suite with environment variable substitution and assertions:

prism run ./tests/api-suite.json

With environment overrides:

prism run ./tests/api-suite.json -e ./tests/production.env.json

Collection Format Example:

{
  "name": "User Service Tests",
  "env": {
    "BASE_URL": "https://api.example.com"
  },
  "requests": [
    {
      "name": "Health check",
      "method": "GET",
      "url": "{{BASE_URL}}/health",
      "expect": {
        "status": 200,
        "maxTime": 500,
        "headers": {
          "content-type": "application/json"
        }
      }
    },
    {
      "name": "Authentication",
      "method": "POST",
      "url": "{{BASE_URL}}/auth/token",
      "body": { "client_id": "test_app" },
      "extract": {
        "AUTH_TOKEN": "token"
      },
      "expect": {
        "status": 200
      }
    },
    {
      "name": "Get Profile",
      "method": "GET",
      "url": "{{BASE_URL}}/profile",
      "headers": {
        "Authorization": "Bearer {{AUTH_TOKEN}}"
      },
      "expect": {
        "status": 200,
        "bodyContains": "username"
      }
    }
  ]
}

3. High-Concurrency Benchmarking

Execute load and stress testing against any target HTTP/HTTPS endpoint:

# 50 concurrent workers for 15 seconds
prism bench https://httpbin.org/get -c 50 -d 15

# Execute exactly 10,000 requests across 100 workers
prism bench https://httpbin.org/get -c 100 -n 10000

Metrics reported:

• Total requests and completion rate
• Throughput (Requests per second & Megabytes per second)
• Latency percentiles: Min, Mean, p50 (Median), p90, p95, p99, Max, StdDev
• HTTP status code breakdown (2xx, 3xx, 4xx, 5xx, errors)

4. Embedded OLED Studio

Launch the local web dashboard:

prism studio

Options:

• -p, --port <port>: Port to listen on (default: 7070)
• -h, --host <host>: Bind host address (default: 127.0.0.1)

Features:

• Interactive Workbench: Request composer supporting methods, parameters, custom headers, JSON/raw payloads, and Bearer/Basic authentication.
• Latency Inspector: Visual cards showing DNS, TCP, TLS, TTFB, and response duration.
• Live Benchmark Runner: Server-Sent Events (SSE) streaming real-time requests/sec, progress percentage, and latency tables.
• Suite Runner: In-browser test runner displaying step-by-step pass/fail assertions.

Programmatic Usage

Prism can be imported into Node.js scripts or CI/CD pipelines:

const { executeRequest, runCollection, BenchEngine } = require('prismbench');

// Execute single request
const response = await executeRequest({
  url: 'https://httpbin.org/get',
  method: 'GET'
});
console.log(response.status, response.timings.total);

// Run load benchmark
const bench = new BenchEngine({
  url: 'https://httpbin.org/get',
  concurrency: 20,
  duration: 5
});
const report = await bench.run();
console.log(`Throughput: ${report.requests.rps} req/sec`);

Quality & Compliance

• Zero Emojis: Strictly pure ASCII/ANSI logging across CLI, UI, and documentation.
• Zero Mock Data: All metrics originate from real operating system socket events and network payloads.
• Deterministic: Standardized exits (0 on pass, 1 on failure) for automated CI integration.

License

MIT License. Copyright (c) 2026 Atlas Research Labs.