gitnexus wiki

Generate LLM-powered documentation from the knowledge graph

The wiki command generates comprehensive documentation for a repository by analyzing the knowledge graph with an LLM.

Usage

gitnexus wiki [path] [options]

Arguments

path · string

Path to the repository to document. Defaults to current directory’s git root.

Options

force · boolean · default: "false"

Force full regeneration even if wiki is up to date.

Flag: -f, --force

model · string · default: "minimax/minimax-m2.5"

LLM model name. Supports OpenAI, OpenRouter, or any OpenAI-compatible API.

Flag: --model <model>

Examples:

• gpt-4o-mini (OpenAI)
• minimax/minimax-m2.5 (OpenRouter)
• llama3 (Local via Ollama)

baseUrl · string · default: "https://openrouter.ai/api/v1"

LLM API base URL.

Flag: --base-url <url>

Providers:

• OpenAI: https://api.openai.com/v1
• OpenRouter: https://openrouter.ai/api/v1
• Ollama: http://localhost:11434/v1

apiKey · string

LLM API key. Saved to ~/.gitnexus/config.json for future use.

Flag: --api-key <key>

Alternatively, set environment variable:

• GITNEXUS_API_KEY
• OPENAI_API_KEY

concurrency · number · default: "3"

Number of parallel LLM calls.

Flag: --concurrency <n>

gist · boolean · default: "false"

Publish the generated wiki as a public GitHub Gist for easy viewing and sharing.

Flag: --gist

Requires: GitHub CLI (gh) installed and authenticated

Examples

Basic Usage

Generate wiki for the current repository:

cd my-project
gitnexus wiki

First Run (Interactive Setup):

  GitNexus Wiki Generator

  No LLM configured. Let's set it up.

  Supports OpenAI, OpenRouter, or any OpenAI-compatible API.

  [1] OpenAI (api.openai.com)
  [2] OpenRouter (openrouter.ai)
  [3] Custom endpoint

  Select provider (1/2/3): 2
  Model (default: minimax/minimax-m2.5): 
  API key: ****************************************
  Config saved to ~/.gitnexus/config.json

Custom Model

gitnexus wiki --model gpt-4o-mini

Local LLM (Ollama)

gitnexus wiki --base-url http://localhost:11434/v1 --model llama3

Force Regeneration

gitnexus wiki --force

Publish as Gist

gitnexus wiki --gist

Save API Key

gitnexus wiki --api-key sk-...

Key is saved to ~/.gitnexus/config.json and reused for future runs.

Output

The command displays a progress bar with phase updates:

  GitNexus Wiki Generator

  ████████████████████░░░░ 85% | Generating modules (24/32)

Successful Generation

  Wiki generated successfully (127.3s)

  Mode: full
  Pages: 32
  Output: /Users/dev/my-project/.gitnexus/wiki
  Viewer: /Users/dev/my-project/.gitnexus/wiki/index.html

  Publish wiki as a GitHub Gist for easy viewing? (Y/n): y

  Publishing to GitHub Gist...
  Gist:   https://gist.github.com/user/abc123
  Viewer: https://gistcdn.githack.com/user/abc123/raw/index.html

Already Up to Date

  Wiki is already up to date.
  Viewer: /Users/dev/my-project/.gitnexus/wiki/index.html

Generated Files

Wiki output is saved to .gitnexus/wiki/:

.gitnexus/wiki/
├── index.html           # Single-file viewer (open in browser)
├── README.md            # Overview
├── modules/
│   ├── authentication.md
│   ├── database.md
│   └── ...
└── architecture.md      # High-level architecture

Viewing the Wiki

Open index.html in your browser:

open .gitnexus/wiki/index.html  # macOS
xdg-open .gitnexus/wiki/index.html  # Linux
start .gitnexus/wiki/index.html  # Windows

How It Works

Phase 1: Graph Analysis (0-20%)

1. Loads the knowledge graph from .gitnexus/db
2. Identifies functional modules (communities)
3. Extracts entry points and execution flows (processes)

Phase 2: Module Generation (20-90%)

For each module:

1. Query the graph for symbols, relationships, and processes
2. Call LLM with structured context
3. Generate documentation (purpose, components, flows)
4. Save markdown to modules/{module}.md

Concurrency: Multiple modules are processed in parallel (default: 3).

Phase 3: Architecture Overview (90-100%)

Generates high-level architecture documentation by synthesizing module relationships.

Phase 4: HTML Viewer Generation

Compiles all markdown files into a single-file HTML viewer with:

• Sidebar navigation
• Full-text search
• Syntax highlighting
• Responsive design

LLM Configuration

Providers

Provider	Base URL	Model Example
OpenAI	https://api.openai.com/v1	gpt-4o-mini
OpenRouter	https://openrouter.ai/api/v1	minimax/minimax-m2.5
Ollama	http://localhost:11434/v1	llama3
Custom	Your URL	Your model

Default Model

minimax/minimax-m2.5 via OpenRouter — fast, cheap, and high-quality for documentation generation.

Cost Estimates

Repository Size	Modules	Tokens	Cost (OpenRouter)
Small (< 100 files)	5-10	50k	$0.05
Medium (100-1,000 files)	10-20	200k	$0.20
Large (1,000-10,000 files)	20-50	500k	$0.50

Costs are approximate. OpenRouter’s minimax/minimax-m2.5 is optimized for cost.

GitHub Gist Publishing

The --gist flag publishes the HTML viewer to a public GitHub Gist.

Requirements

1. GitHub CLI (gh) installed: https://cli.github.com
2. Authenticated: gh auth login

Automatic Publishing

If --gist is omitted, you’ll be prompted after generation:

Publish wiki as a GitHub Gist for easy viewing? (Y/n): 

Output

Gist:   https://gist.github.com/user/abc123
Viewer: https://gistcdn.githack.com/user/abc123/raw/index.html

Share the Viewer URL for instant access (no GitHub login required).

Caching and Incremental Updates

Wiki generation tracks the last indexed commit. If the repository hasn’t changed:

Wiki is already up to date.

Use --force to regenerate:

gitnexus wiki --force

Troubleshooting

”No GitNexus index found”

Run gitnexus analyze first:

gitnexus analyze
gitnexus wiki

“No LLM API key found”

Provide an API key:

gitnexus wiki --api-key sk-...

Or set an environment variable:

export GITNEXUS_API_KEY=sk-...
gitnexus wiki

LLM API Error (401 Unauthorized)

Your API key is invalid. Reconfigure:

gitnexus wiki --api-key sk-new-key

Failed Modules

Failed modules (3):
  - authentication
  - billing
  - notifications
Re-run to retry failed modules (pages will be regenerated).

Solution:

gitnexus wiki  # Retries only failed modules

Gist Publishing Fails

Ensure GitHub CLI is authenticated:

gh auth status
gh auth login  # If not authenticated

Performance

Repository Size	Modules	Time (concurrency=3)
Small (< 100 files)	5-10	30-60s
Medium (100-1,000 files)	10-20	1-3 min
Large (1,000-10,000 files)	20-50	3-10 min

Increase concurrency for faster generation:

gitnexus wiki --concurrency 10

Higher concurrency = higher LLM costs and rate limit risk.

See Also

• gitnexus analyze — Index repository first
• Web UI Overview — Alternative documentation viewer