THIS PROJECT IS UNMAINTAINED AND SOME VULNERABILITIES HAS BEEN FOUND IN THE SUPPLY CHAIN.
- Uncaught Exception in jsoup < 1.14.2.
- Path Traversal and Improper Input Validation in Apache Commons IO < 2.7.
- TemporaryFolder on unix-like systems does not limit access to created files >= 4.7, < 4.13.1.
This is a stable working Alexa Skill source code with a deployment help script used for a former unsuccessful business.
After I attended two workshops about AWS Alexa and I spent a few hours reading articles about how to program a Skill, I really believed that programming a Skill I could start a plenty unattended business, something like dreaming that anyone can be rich programming a mobile App. But in this case, voice services are quite new and there is less competitors involved. I was wrong in many ways.
I spent many days working on that Skill. Alexa is going to read the website while you are driving, or just resting. It was a huge work from BDD, and finally I released my first version.
Oh... one of the main Alexa Skill policies tells that an opened websearch engine is prohibited. I had to search only within a previously declared domains. That made me spend a few hours more.
Finally I released and I got the skill certified limiting the search queries within one only domain, a very popular publishing platform. I released for spanish first and english later. Spanish certification was quite easier than english one, much much more.
I got a mean of 2 or 3 users every week, and after a few months I only got two user feedback, one worse and the other worst.
Meanwhile the Skill was in production, I decided to deepen into AWS Alexa Skills. Thus I realized that my skill approach is totally wrong. This skill works like an automated phone service, and that's what Alexa guidelines want you keep out of your mind when you are designing a skill.
I recommend to carefully follow ALL the "getting started" tutorials provided by AWS before your first important skill, otherwise you are going to lose your valued time.
I made public this project because despite the Alexa Skill approach is totally wrong, the architecture is highly maintainable, and may be one of you want to resume the idea where I left it.
There are many abstractions where you can interfere with new implementations according a new approach. Actually the wrong part is isolated along the AWS implementation and infrastructure layers only.
- AWS Account and a user with programmatic access to use the deployment script.
- AWS IAM Role with permissions to execute Lambda Functions and handling DynamoDB records.
- An Alexa Skill created through the Alexa Skill console.
Deployment script consist on an AWS Lambda function update and remote test and a DynamoDB tables creation command.
You have to provide an AWS account, a role allowed to execute functions and handle DynamoDB records. That's not a magic installation and deployment script. It helped me a lot because upgrading the AWS Lambda by hand, in the console, is quite tedious.
- /deploy/conf/global.sh.template: copy this file into a new global.sh file and configure the parameters before deployment execution.
- /deploy/deploy_lambda.sh: Deployment script for building, uploading and configuring the AWS Lambda function.
AWS Alexa Skill model.
Java 8 source code with maven POM file dependencies management.
- /src/main/java/com/westial/alexa/jumpandread: production source code.
- /src/test/features: behavioural gherkin tests with no integration.
- /src/test/java/junit: JUNIT based integration tests.