ARCHITECTURAL WALKTHROUGH

How PerformancePlatform Works

Understand the complete workflow from organization onboarding to local k6 load generation and AI-assisted performance engineering.

Step 1: Onboarding

Register & Access Your Tenant Workspace

Sign up with your team email to create an isolated organization tenant. Every new organization immediately receives a 3-execution lifetime Free Trial with no credit card required.

POST /api/auth/register → 200 OK [JWT issued with org_id & trial subscription]

Step 2: Projects

Create or Select a Testing Project

Projects group related API services, test definitions, baseline runs, and diagnostic investigation records. Multiple team members can collaborate within the same organization.

Organize by microservice, staging environment, or customer domain.

Step 3: Cataloging

Import OpenAPI Specification or Add Endpoints

Upload a Swagger or OpenAPI 3.0 JSON specification to auto-populate your API catalog. Endpoints, query parameters, headers, and request bodies are extracted automatically.

Attach Authentication Profiles: configure Bearer tokens, API keys, or Basic credentials securely.

Step 4: Configuration

Configure Test Parameters & Validate Readiness

Set target URLs, Virtual Users (VUs), duration, ramp-up curve, and synthetic test data generators. The built-in Readiness Panel checks for schema mismatches and agent connectivity before starting.

Test Parameters: 100 VUs • 30s Duration • 5s Ramp-up • Synthetic Dynamic Payloads

Step 5: Agent Connectivity

Connect Customer-Side k6 Agent

Generate an agent registration token in the dashboard, download the standalone agent package, and launch it inside your network. It establishes an outbound encrypted SignalR WebSocket to the SaaS hub.

$ $env:AGENT_TOKEN = "agt_tok_..."
$ .\PerformancePlatform.Agent.exe

Step 6: Execution

Execute Load Test Using k6 in Your Environment

When dispatched, the agent receives the test configuration, generates synthetic payloads, runs k6 as an isolated subprocess, and streams execution metrics back in real time.

Target services are contacted locally within your VPC—traffic never routes through public internet proxies.

Step 7: Analysis

Review Metrics, Compare Runs & Evaluate SLO Targets

Inspect comprehensive latency histograms: average, P50, P90, P95, and P99. Compare the execution to an approved baseline run to quantify regressions and check SLO compliance.

Deterministic Pass/Fail evaluation against configured performance criteria.

Step 8: AI Workflows

Prompt Compatible AI Clients via MCP

Optionally connect Cursor, Claude Desktop, or VS Code using the Model Context Protocol. Prompt your AI directly: "Run load test on checkout and analyze tail latency regressions."

Discovered Tools: list_agents, create_performance_test, run_load_test, compare_test_runs, evaluate_performance_targets

Ready to Run Your First Load Test?

Get Started Free