Krkn

Term 2 · Jun-Aug · 2 projects

• Automated Documentation Sync Bot for Krkn-Chaos Projects

  LFX Mentorship · Listed in the official cohort

  @rh-rahulshettySource IssueLFX portal

  The krkn-chaos/website repository hosts the unified documentation for the entire krkn-chaos ecosystem (krkn, krkn-hub, krknctl, krkn-ai, krkn-operator, cerberus, etc.) as a Hugo/Docsy site deployed at krkn-chaos.dev. Currently, when changes are made to any of these upstream projects — such as adding a new scenario, modifying configuration options, or updating CLI flags — someone must manually open a PR against the website repo to update the corresponding docs. This is tedious, easy to forget, and often leads to documentation drift where the docs no longer match the actual tool behavior. This issue proposes building a bot or GitHub Action-based workflow that detects documentation-impacting changes in upstream krkn-chaos repositories and automatically creates a draft PR on the website repo with the relevant documentation updates. The bot would analyze the scope of the change (e.g., new config fields, updated parameters, new scenario) and generate or update the corresponding Hugo content files, following the existing content structure and conventions (tabbed scenario layouts, parameter tables, etc.).

  • Python
  • GitHub Actions/Workflows
  • Hugo static site basics
  • LLM
  • Agent development

• Dynamic Cluster-Aware Configuration Generation for Krkn-AI

  LFX Mentorship · Listed in the official cohort

  @rh-rahulshettySource IssueLFX portal

  Krkn-AI's discover command connects to a Kubernetes/OpenShift cluster and generates a static configuration file by enumerating cluster components (namespaces, pods, services, PVCs, nodes). While useful, the generated config still requires significant manual work before it can actually be used. Health check URLs are commented-out placeholders, the fitness function defaults to a single hardcoded PromQL query, and scenario selection is static regardless of what exists in the cluster. This issue proposes enhancing discover to produce a dynamic, cluster-aware configuration that is closer to runnable out-of-the-box. By inspecting routes, ingresses, services, and available Prometheus metrics during discovery, we can auto-populate health check URLs, suggest relevant fitness function queries scoped to discovered namespaces, and intelligently enable only the scenarios that apply to the discovered infrastructure.

  • Python
  • Kubernetes/OpenShift (client libraries
  • API concepts like Routes
  • Ingresses
  • Services)
  • Prometheus/PromQL basics

Term 1 · Mar-May · 2 projects

• Enhancing Krkn-AI Result Analysis with Interactive Visualization and Insights

  LFX Mentorship · Listed in the official cohort

  @rh-rahulshettySource IssueLFX portal

  Krkn-AI generates rich but complex experiment outputs (JSON, CSV, YAML, graphs, and tables) capturing fitness scores, SLOs, health checks, and other metrics, which can be difficult for engineers to interpret and compare across experiments. Although recent work using LLMs to produce high-level textual summaries is helpful, text alone limits deeper exploration. This feature change proposes building an interactive analysis and visualization layer for Krkn-AI that transforms raw chaos experiment data into intuitive, explorable visual representations, enabling users to quickly understand system behavior, detect anomalies, and focus on the most impactful failure signals.

  • Python
  • Data Analysis and Visualization tools (e.g: pandas
  • matplotlib)
  • Kubernetes (Basics)

• Natural Language–Based Chaos Scenario Discovery

  LFX Mentorship · Listed in the official cohort

  @paigerube14Source IssueLFX portal

  The objective of this internship task is to design and implement a feature of the [krknctl]( tool that enables users to explore the project’s functionality by submitting queries in natural language. The tool will analyze the user’s request and Identify the most relevant chaos scenario, if any.

  • Python and Go programming
  • Basic knowledge of Natural Language Processing (NLP)
  • Experience with machine learning models for text classification or similarity
  • Familiarity with Docker and containerized applications
  • Ability to write clean
  • testable
  • +1

Term 3 · Sep-Nov · 1 project

• Implementing the resiliency score feature

  LFX Mentorship · Listed in the official cohort

  @tsebastianiSource IssueLFX portal

  The idea is to define criteria to measure the resiliency score of a cluster after a set of chaos scenarios.

Term 2 · Jun-Aug · 1 project

• Chaos scenario rollback feature

  LFX Mentorship · Listed in the official cohort

  @tsebastianiSource IssueLFX portal

  <br/> We intend to refactor the current Scenario Plugin API and add a new feature in the plugin abstract class to restore all the changes performed by krkn in the cluster in case run method encounters some kind of exception before ending the execution and or if the run failed for external infrastructure failures.

  • <br/>
  • python
  • kubernetes
  • container runtime environments
  • common cloud platforms