Open Source Friday with Keploy - AI-Powered Testing for Developers!

Key Concepts

• GitHub Copilot: An AI pair programmer that assists developers with coding tasks.
• Copilot Chat: An interface within GitHub where users can interact with Copilot using natural language.
• GitHub Issues: A system for tracking tasks, bugs, and feature requests within a GitHub repository.
• Model Context Protocol (MCP): A protocol that allows Copilot to access external tools and data sources, such as Notion.
• GitHub Actions: A platform for automating software workflows, such as building, testing, and deploying code.
• Kiplo: An open-source tool for automated API testing with AI.
• EBPF (Extended Berkeley Packet Filter): A technology used by Kiplo to intercept network calls at the kernel level.
• API Testing: Testing the functionality, reliability, and performance of APIs.
• Test Cases: A set of conditions or variables under which a tester will determine whether an application, software system or one of its features is working as it was designed.
• Test Suite: A collection of test cases that are intended to be used to test a software program to show that it has some specified set of behaviors.

GitHub Copilot and Issue Creation

• Copilot Chat allows users to create GitHub issues directly from the chat interface.
• Example: A product team using Copilot Airways wants to create an issue for user reviews on the travel guide, which is documented in Notion.
• Copilot can create a GitHub issue with a link to the Notion document, providing traceability for the engineering team.
• Copilot coding agent can be assigned issues directly from Copilot Chat.

Model Context Protocol (MCP)

• MCP enables Copilot to access external tools and data sources.
• Example: Copilot accesses product documentation in Notion using MCP.
• Copilot uses multiple tool calls to gather information from Notion and incorporate it into its workflow.
• MCP servers and associated tools are configured during Copilot setup in GitHub Actions.
• MCP.json file configures the MCP servers that are available.
• Different server types: stdio serverance events (SSE).
• Stdio can run npm packages, pip packages, docker images, or local commands.
• Requires authentication with a personal access token (PAT) with specific permissions.
• PAT should not be stored in source code.

Copilot and Code Review

• Copilot can summarize changes made in a pull request, providing context for code review.
• This helps reviewers understand the changes and provide more effective feedback.

Building a Front-End Project with Copilot

• Copilot can assist in building front-end projects, such as React clients.
• Copilot Chat can provide detailed descriptions of the code in a directory.
• Copilot can edit multiple files at once, allowing for iterative development.
• Developers can provide additional suggestions to Copilot, refining the code.
• Example: Building a fully functional MVP of an application in a short time with Copilot.

Pushing to GitHub with Copilot

• Copilot can push repositories to GitHub, including creating new repositories.
• This functionality is enabled by MCP support.
• MCP allows the agent to use tools to interact with GitHub.

Open Source Friday and Kiplo

• Open Source Friday is a show where maintainers discuss their open-source projects.
• Niha, a core maintainer of Kiplo, discusses the tool.
• Kiplo is an agent API platform based on EBPF that intercepts network calls and converts API traffic to test cases and data mocks.
• Kiplo aims to provide good test coverage without requiring developers to spend excessive time writing test cases.
• Kiplo was created out of frustration with the time spent fixing or writing test cases.

Kiplo vs. Traditional Testing Tools

• Traditional record and playback tools require manual preparation of mocks and ensuring the test environment is in the same state.
• Kiplo records database calls and external service interactions, automating the creation of mocks.
• Kiplo records the database call as well while recording.

Kiplo Demo

• Niha demonstrates Kiplo by recording API test cases for GitHub.
• She opens github.com, records some calls, and copies the interactions.
• She uses these curls as input, along with the open API schema of GitHub.
• Kiplo generates test cases based on the schema and recorded interactions.
• Example test case: Create a repository, check if the API call returns 201, create an issue, get an issue, and clean up the repository.
• Kiplo generates multiple edge cases for the GitHub API.
• Test cases can be run whenever there is a new release.
• Kiplo can be integrated with GitHub Actions.

Kiplo Configuration

• Configuring Kiplo in a repository is simple, requiring adding an action to the workflow file in GitHub repository.
• For developers who prefer to record their own flows, Kiplo can wrap a JAR file and record API calls and responses.
• Kiplo also records mocks, eliminating the need to set up databases.
• Test cases can be run without a database.

Kiplo and Complex Offflows

• Kiplo handles complex offflows by recording large payloads and allowing interchaining of APIs.
• The response of one API can be used as the request of another.

Kiplo Language Support

• Kiplo is language agnostic.
• The underlying LLMs can be chosen by developers.
• Kiplo focuses on the types of databases or protocols being used.
• Currently supports popular databases like MongoDB, Postgress, and MySQL.

Kiplo CI/CD Integration

• Kiplo can be integrated with any CI/CD platform where a binary can be downloaded and EBPF hooks can be plugged in.
• It is commonly used with GitHub Actions, GitLab, and Bitbucket.

Kiplo API Traffic Capture

• Kiplo captures real-time API traffic by monitoring network calls.
• It uses a Chrome extension to see what's happening behind the scenes of the application.
• It uses all those events to automatically generate test cases.

Kiplo Tech Stack

• Kiplo is built in Golang.
• It uses EBPF to intercept network calls at the kernel layer.
• This eliminates the need for code integration or SDKs.

Kiplo and Evolving APIs

• Kiplo monitors APIs and generates test cases for new APIs.
• If something is breaking due to schema changes, Kiplo can self-heal the test case using AI.

Kiplo Platform Overview

• The Kiplo platform allows users to share test suites with colleagues, view reports, and add resources.
• It provides a dashboard to visualize which endpoints have been covered from the schema.
• It is designed to be easy to get up and running within a few minutes.

Challenges in Building Kiplo

• The most challenging part was mapping API requests to the dependencies of the application without injecting any SDK.
• This required writing protocol parsers, which did not exist.
• Building a proxy and parsers for the database was difficult and time-consuming.

Kiplo and Community

• Kiplo has received a lot of help from the community.
• Users have provided feedback and helped identify failing points.
• The community has also contributed to new features, such as gRPC support.

Kiplo and Open Source Contributions

• Kiplo accepts open-source contributions.
• There are over 100 issues on the repository.
• Contributions can include adding new databases, language frameworks, and reporting coverage libraries.
• Kiplo also accepts contributions from non-coding backgrounds, such as blogs or tutorials.

Kiplo and Dynamic Values

• Kiplo handles dynamic or non-deterministic values, such as UYU ids or timestamps, by adding them as noise.
• Noisy fields are not compared in the actual and expected responses.
• Wildcard entries can be added to ignore fields completely.

Conclusion

• GitHub Copilot and Kiplo are powerful tools that can help developers improve their productivity and code quality.
• Copilot assists with coding tasks, issue creation, and code review.
• Kiplo automates API testing, reducing the time and effort required to write and maintain test cases.
• Both tools are designed to be easy to use and integrate with existing workflows.