NANY 2026: Akari Flow - AutoDocumentation (and more) CI/CD for GitHub

[KynloStephen66515]

Repo: https://github.com/Kynlos/Akari-Flow

Akari Flow

Intelligent CI/CD Documentation & Analysis System

Why Akari Flow? 
"Documentation is the code." — Everyone says it, but nobody actually searches the code. In reality, documentation rots the moment it's written. Developers hate writing it, and teams struggle to maintain it. The cost? Slower onboarding, silent breaking changes, and critical knowledge locked in people's heads. 
This system aims to eliminate documentation debt entirely.

Overview 
Akari Flow revolutionizes code documentation by automatically analyzing code changes with AST-based precision, creating diagrams, generating docs, detecting breaking changes, and notifying teams.

✨ Key Features 

• Automated Code Analysis – Breaking change detection, impact analysis, quality scores, security scanning. 
  
• Intelligent Documentation & Visualization – Auto docs, architecture diagrams, static site generation. 
  
• PR Enhancement & Agentic Bot – Smart PR comments and safe search/replace automation. 
  
• Multi-Platform Notifications – Discord, Slack, Pushbullet integration. 
  

Quick Start 
Prerequisites: 

• GitHub repository 
• Groq API key (free tier available) 
• Python 3.11+ 

5-Minute Setup: 

[Select]

git clone https://github.com/Kynlos/Akari-Flow.git
cd Akari-Flow

# Copy workflows into your repo
cp -r .github <your-repo>/

# Add required secret
# GROQ_API_KEY = <your-groq-api-key>

Push a change and the magic starts.

CLI Tool (Local Usage) 
Install dependencies and run:

[Select]

pip install requests markdown

python cicd-cli.py full-run

Generate docs or site with dedicated commands.

Configuration 
Language parsing rules and LLM model settings are fully configurable.

Usage Examples 
Agentic Bot – Comment on PRs to modify code safely. 
Breaking Change Detection – Signature changes trigger alerts.

Documentation Portal 
The system builds a static site with navigation, dark mode, and diagrams.

Architecture 
The pipeline runs as GitHub Actions — parse code, generate docs & diagrams, build site, send notifications, then commit back.

Support 
If you find this project useful, consider sponsoring development.

About 
Intelligent CI/CD: Automated documentation, AST-based analysis, breaking change detection, and dynamic visualization powered by AI.

[Shades]

Looks interesting, so thanks 

As I have been playing around with the latest (and largest that fits) Mistral LLM in my local setup, it produces quite useful results with code. That gives me the impression that it would also do great with generating documentation from code.

Instead of Grok, can I use local LLMs instead?

[KynloStephen66515]

Looks interesting, so thanks 

As I have been playing around with the latest (and largest that fits) Mistral LLM in my local setup, it produces quite useful results with code. That gives me the impression that it would also do great with generating documentation from code.

Instead of Grok, can I use local LLMs instead?

-Shades (December 24, 2025, 05:56 PM)

Groq* (Grok is X.com, groq is a different provider)

That aside, it wouldn't be hard for me to add this, however, I'm not sure how it would work given you'd need to have your local LLM publicly available in order for GitHub to reach it and do stuff

What sort of workflow are you thinking?

[KynloStephen66515]

OK, now it defaults to Groq if you don't have any specifically set, but you can do it in Git Secrets if you want to set an Ollama BASE_URL for example:

[Select]

# 3. Configure GitHub Settings > Secrets and variables > Actions
#   - New Repository Secret: GROQ_API_KEY (or LLM_API_KEY)
#   - New Repository Variable: LLM_BASE_URL (Optional, for custom providers)
#   - New Repository Variable: LLM_MODEL (Optional, e.g. "llama3")

API key is optional if you don't have one for your locally hosted models