avdoc

AI-powered documentation gatekeeper and architecture visualizer.

Overview

avdoc provides tools to analyze and improve repository documentation. It can evaluate codebases to generate a documentation quality score, automatically generate missing documentation, and produce architecture diagrams.

Installation

Option 1 — cargo install (recommended for Rust users)

cargo install avdoc

Option 2 — One-line install script (Mac / Linux)

curl -sSf https://raw.githubusercontent.com/Aryan-202/avdoc/main/install.sh | sh

Option 3 — One-line install script (Windows PowerShell)

irm https://raw.githubusercontent.com/Aryan-202/avdoc/main/install.ps1 | iex

Option 4 — Manual download

Download the pre-built binary for your platform from the latest release:

Platform	Binary
Linux x86_64	avdoc-linux-x86_64
Linux ARM64	avdoc-linux-arm64
macOS x86_64	avdoc-macos-x86_64
macOS ARM64 (Apple Silicon)	avdoc-macos-arm64
Windows x86_64	avdoc-windows-x86_64.exe

Requirements

• An API key from OpenAI, Anthropic, or a compatible provider
• Set one of the following environment variables before use:

export OPENAI_API_KEY=your_key_here
# or
export ANTHROPIC_API_KEY=your_key_here
# or
export AVDOC_API_KEY=your_key_here

To use a specific model, set:

export AVDOC_MODEL=gpt-4o   # defaults to gpt-3.5-turbo

---

Commands

avdoc lint

Lint the repository to evaluate documentation coverage and quality.

avdoc lint [OPTIONS]

Option	Description
--path <DIRECTORY>	Path to the repository (defaults to current directory)
--min-score <SCORE>	Enforce a minimum documentation score from 0 to 100
--format <FORMAT>	Output format: terminal, json, markdown

Examples:

# Lint current directory
avdoc lint

# Fail CI if score drops below 80
avdoc lint --min-score 80

# Output as JSON
avdoc lint --format json

# Lint a specific repo path
avdoc lint --path ./my-project

---

avdoc diagram

Generate architecture diagrams and visually map the repository structure.

avdoc diagram [OPTIONS]

Option	Description
--path <DIRECTORY>	Path to the repository
--format <FORMAT>	Output format: mermaid (default), ascii
--update-readme	Automatically append the diagram to README.md

Examples:

# Generate a Mermaid diagram
avdoc diagram

# Generate ASCII diagram
avdoc diagram --format ascii

# Generate and append to README
avdoc diagram --update-readme

---

avdoc heal

Analyze and automatically generate missing documentation for the codebase using AI.

avdoc heal [OPTIONS]

Option	Description
--path <DIRECTORY>	Path to the repository
--files <FILES>...	Target specific files; if omitted, targets all low-scoring files
--interactive	Ask for confirmation before applying each change

Examples:

# Heal all underdocumented files
avdoc heal

# Heal specific files
avdoc heal --files src/main.rs src/lib.rs

# Review each change before applying
avdoc heal --interactive

---

Updating

Re-run your original install command to get the latest version. If you installed via cargo:

cargo install avdoc

avdoc follows semantic versioning: patch versions (0.1.x) are bug fixes, minor versions (0.x.0) add new features, and major versions (x.0.0) indicate breaking changes.

---

Supported Languages

avdoc currently parses and scores documentation for:

• Rust
• Python
• JavaScript / TypeScript
• Go
• Java
• C / C++

---

CI Integration

You can use avdoc as a documentation gate in CI pipelines. It exits with a non-zero code if the score falls below the minimum:

# GitHub Actions example
- name: Check documentation score
  run: avdoc lint --min-score 70
  env:
    OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}

---

License

MIT License — see LICENSE for details.