Why it is worth attention

It combines natural-language Gherkin, managed cloud browsers, parallel execution, and end-to-end observability in a ready-to-run demo, including a live-view webapp.

Who it is for

QA engineers evaluating AI-driven end-to-end testing
Developers working with Amazon Bedrock AgentCore Browser
Teams using or evaluating Langfuse/OpenTelemetry for LLM tracing
Technical speakers or demo builders needing an AI QA example

Use cases

Run Gherkin scenarios against a real website without CSS/XPath selectors
Parallel smoke tests with configurable browser-session concurrency
Observe agent model calls, costs, and browser actions in Langfuse
Give a live demo with an AWS Console live browser view

Strengths

Supports plain-language Gherkin features in Spanish or English; no selectors or Playwright code needed
Runs scenarios concurrently in separate cloud browser sessions
Traces every scenario as a Langfuse session/trace with model calls, tokens, cost, and browser actions
Offline tests are fully mocked and maintain 100% coverage with lint/type checks

Considerations

Requires Python 3.12+, uv, AWS credentials with specific Bedrock AgentCore permissions, and Bedrock model access
Live-view feature depends on extra AWS permission and signed URLs that expire after five minutes
It is a demo-level project, not positioned as a full production test framework

README quick start

QA Agent — Strands Agents + AgentCore Browser + Langfuse

Demo de QA automático con agentes: un agente construido con Strands Agents SDK ejecuta casos de prueba escritos en Gherkin contra un sitio real, usando el browser gestionado de Amazon Bedrock AgentCore. Cada escenario corre en paralelo en su propia sesión de browser en la nube, y toda la ejecución queda trazada en Langfuse vía OpenTelemetry.

Gherkin (.feature) ──► Runner (asyncio, N en paralelo)
                            │  un agente Strands por escenario
                            ▼
                     AgentCore Browser (Chrome gestionado en AWS)
                            │  navega, hace clic, lee la página
                            ▼
                     Veredicto estructurado (PASS/FAIL + evidencia)
                            │
                            ▼            trazas OTel
                     Reporte en consola      ──►  Langfuse

Requisitos

Python 3.12+ y uv
Credenciales AWS configuradas (aws sts get-caller-identity) con permisos bedrock-agentcore:*BrowserSession* y bedrock:InvokeModel*
Model access a Claude habilitado en la consola de Bedrock
Cuenta en Langfuse Cloud (gratis) con API keys del proyecto

Setup

uv sync
cp .env.example .env   # y llena las keys de Langfuse

Uso

# Listar escenarios del feature
uv run qa-agent features/tuasistente.feature --list

# Ver los prompts sin llamar al agente
uv run qa-agent features/tuasistente.feature --dry-run

# Correr UN escenario
uv run qa-agent features/tuasistente.feature -s "La home carga con la propuesta de valor"

# Correr TODOS en paralelo (3 sesiones de browser concurrentes por defecto)
uv run qa-agent features/tuasistente.feature

# Subir la concurrencia
uv run qa-agent features/tuasistente.feature --parallel 5

Exit code 0 si todo pasó, 1 si hubo fallos — listo para CI.

Webapp con Live View

La demo también se puede ejecutar desde una interfaz local que recibe el feature Gherkin y muestra una tarjeta por escenario con el navegador de AgentCore en vivo:

# Terminal 1: API que lanza los agentes
uv run qa-agent-web

# Terminal 2, sólo para desarrollar la interfaz (abre http:

Related repositories

Similar projects matched by category, topics, and programming language.

lopopolo

Featured

harness-engineering

Harness Engineering is a methodology for improving coding agent outputs by carefully crafting the environment around them—providing curated context, tools, and executable constraints that encode an organization’s nonfunctional requirements and cumulative lessons.

AI & Machine LearningAI Agents

2,390

slvDev

Featured

esp32-ai

A 28.9 million parameter language model runs on an $8 ESP32-S3 microcontroller entirely on-device, generating simple stories at about 9.5 tokens per second.

AI & Machine LearningLarge Language Models

1,960

littledivy

Featured

mimic

mimic captures traffic from any iOS or web app and automatically generates a Python client library that lets you call the app's API like a regular library.

AI & Machine Learning

1,482