Loadforge

A high-performance API load testing application with a Python async engine, real-time analytics, and a modern React dashboard.

Browser (React + Recharts)
    │ SSE
    ▼
Vite (:5173) ──proxy──▶ Node/Express (:3001) ──SSE proxy──▶ Python/FastAPI (:3002)
                              │                                     │
                              │ history, config                     │ aiohttp + asyncio
                              ▼                                     ▼
                         JSON file storage                    Target Endpoint

The stress engine runs in Python using aiohttp for lower per-request overhead, accurate timing, and better concurrency scaling compared to Node-based alternatives. The Node server handles the API layer, history persistence, and SSE streaming to the browser.

• Target URL with HTTP method selection (GET, POST, PUT, DELETE, PATCH, HEAD)
• Custom request headers (key-value editor)
• JSON request body editor for POST/PUT/PATCH
• Environment variable substitution using {{variable}} syntax in URLs, headers, and body

• Bearer Token — auto-injects Authorization: Bearer <token>
• Basic Auth — username and password
• API Key — send as header or query parameter, configurable key name

• Total request count (up to 50,000)
• Concurrency level (up to 500 parallel connections)
• Ramp-up time (gradually increase load over N seconds)
• Duration-based mode (run for a fixed time instead of fixed count)
• Request delay (throttle between requests)
• Per-request timeout
• Quick presets: Quick Check, Light, Medium, Heavy, Spike, Endurance

• Configurable warm-up requests that execute before the measured test
• Warm-up results are excluded from all metrics, charts, and the final report

• Status code checks (=, !=, <, >, etc.)
• Response time thresholds
• Body contains text matching
• JSON path extraction and comparison ($.data.id)
• Header exists and header equals checks
• Failed assertions count as test failures
• Per-assertion pass rate tracked and displayed in the report

• Max average response time
• Max P95 / P99 response time
• Max error rate (%)
• Min throughput (req/s)
• Max absolute response time
• Pass/fail verdict displayed in the report with per-check breakdown

• Chain multiple API calls into a user journey
• Each step has its own URL, method, headers, body, and assertions
• Variable extractors — pull values from a response (JSON path, header, or status code) and use them as {{variables}} in later steps
• Per-step delay configuration
• Reorder, add, and remove steps in the UI
• Per-step metrics in the final report (count, avg response, error rate)

• Live-updating stat cards: requests sent, throughput, avg latency, error rate, assertion pass rate
• Status code badges with color coding
• Four streaming charts (Recharts):
  • Throughput over time
  • Response time over time
  • Error rate over time
  • Completed requests over time
• Abort button to stop a running test

Tabbed report with six sections:

• Overview — percentile bar chart, latency histogram, status code pie chart, response time table, bottleneck analysis
• Assertions — per-assertion pass rate with progress bars
• SLA — pass/fail per check with target vs actual
• Timeline — scatter plot waterfall of every request (start offset vs duration)
• Steps — per-step metrics for scenario tests
• Errors — error list with codes and messages

Automatic grading based on response times and error rate:

Automated analysis flags:

• High tail latency (P99 > 3x average)
• High error rates
• Requests approaching timeout
• Throughput below concurrency level
• High response time variance (P90 > 2x average)

• JSON — full test result
• CSV — summary metrics, status codes, histogram, assertions, SLA
• PDF — formatted report with tables (jsPDF + autotable)

• Save test results with custom names
• View past results with the full report
• Delete old entries
• Side-by-side comparison of up to 3 tests (throughput, latency, error rate, grade)
• Persisted to a JSON file on the server (last 100 results)

• Save and reload test configs from the browser (localStorage)
• Includes all settings: URL, auth, assertions, SLA, env vars, load params

• Warning banner about testing only permitted endpoints
• Confirmation dialog for high-load tests (>1000 requests or >100 concurrency)
• Server-side rate limit: max 5 concurrent tests
• Request validation on both client and server

# Install everything
npm run install:all

# Start all three servers
npm run dev

Or individually:

cd worker && python main.py      # Python engine on :3002
cd server && npm run dev          # Node API on :3001
cd client && npm run dev          # React UI on :5173

Open http://localhost:5173 in your browser.

• Node.js 20+
• Python 3.12+
• npm