Docgen

Solve documentation hell for good.

Docgen is a lightweight, C++ CLI tool that automates software documentation using Large Language Models (LLMs). It treats documentation generation as a build step, tracking your source code and incrementally updating documentation only when files change.

Why this solves "Documentation Hell"?

Because you no longer have to bother writing docs yourself, just run docgen update and you are good to go.

Features

• Docs-as-Code: Generates Markdown documentation that lives right next to your source code.
• Smart Incremental Builds: Uses content hashing to only regenerate docs for modified files, saving time and API credits.
• Context-Aware (RAG): Automatically analyzes dependencies to provide relevant context to the AI, ensuring accurate documentation for complex projects.
• Flexible Backend:
  • Cloud: Supports OpenAI, Google Gemini, and compatible APIs.
  • Local: Works fully offline with Ollama (default).
• Zero Dependencies: Compiled as a single static binary.

Getting started

Quick Install

Windows (PowerShell)

irm https://raw.githubusercontent.com/alonsovm44/docgen/master/installer.ps1 | iex

Linux / Mac

curl -fsSL https://raw.githubusercontent.com/alonsovm44/docgen/master/installer.sh | bash

After installation run

docgen init

This creates your Docfile and your .docgen repo. Using the Docfile is easy. Run docgen track <path> this adds the file/directory to your Docfile. Then run docgen update to generate your tracked files documentation. Once generated only files with changes will be updated. You can also run docgen ignore to indicate to Docgen which files or directories to ignore.

Example Docfile

Track:
# This is a comment
    main.cpp
    src/
Ignore:
    src/secret.cpp

Style:
    # these get passed to the AI as parameters when making docs
    -dont use emojis
    -be professional
    -be concise

Other commands

Status

Check which files have changed since the last update.

docgen status

Summary

Generate a SUMMARY.md file with a high-level overview of your project.

docgen summary

Graph

Generate a dependency graph in DOT format (Graphviz) at .docgen/graph.dot.

docgen graph

Clean

Remove documentation for files that are no longer tracked.

docgen clean

Validate

Verify if documentation is up-to-date (useful for CI/CD).

docgen validate

Upgrade

Update docgen to the latest version.

docgen upgrade

Reboot

Run this command to reset your documentation repository (asks for confirmation)

docgen reboot

config

Docgen comes with a config.json file within your ./docgen folder. In it you can configure A.I parameters

You can also run

$ docgen config

Usage: docgen config <key> <value>
Keys: mode, protocol, key, model

Check Connection

To verify if Docgen can connect to the configured AI provider:

docgen config check

Switching Modes

Docgen supports two modes: local (default, uses Ollama) and cloud (uses external APIs).

1. Local Mode (Ollama) This is the default. It requires Ollama running locally.

# Set mode to local
docgen config mode local

# Set the model (e.g., qwen2.5-coder:7b, llama3, mistral)
docgen config model qwen2.5-coder:7b

2. Cloud Mode (OpenAI, Google, etc.) Use this to connect to OpenAI, Google Gemini, or compatible APIs.

# Set mode to cloud
docgen config mode cloud

# Set your API key
docgen config key sk-your-api-key-here

# Set the model ID (e.g., gpt-4o, gemini-1.5-flash)
docgen config model gpt-4o

# Set the protocol (openai, google, or simple)
# 'openai' is standard for most providers including DeepSeek, Groq, etc.
docgen config protocol openai

Configuration Keys reference

Key	Description	Values
mode	Determines the backend to use.	local, cloud
key	API Key for cloud providers.	Your API Key
model	The model identifier to use.	e.g., gpt-4o, llama3
protocol	The API format to use.	openai, google, simple