Gendox — End documentation drift for good

Basic Workflow	Interactive Chat	Auto-Update

Why Gendox matters to YOU NOW.

Docs rot fast, and updating them manually kills focus.
Run gendox auto — your documentation evolves with your code in real time, always accurate, never stale.

Try it now (safe copy-paste)

Note: I get that you may be uncomfortable with piping unknown remote scripts into a shell, you can audit the installer installer.sh/.ps1 yourself and see it is safe. Otherwise you can use shellcheck installer.sh.

Try Gendox in 1 Minute (Instant Onboarding)

Docker (Recommended)

No installation required — just clone and run:

git clone https://github.com/alonsovm44/gendox.git
cd gendox
docker compose up -d                 # Builds Gendox + Ollama
docker compose run gendox init       # Initialize project and makes a .gendox doc repo
                                     # Be sure to fill your Docfile
docker compose run gendox update     # Generate docs

Quick Install (recommended):

Linux / macOS: Download installer, inspect, run:

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

For windows

#download and install
irm https://raw.githubusercontent.com/alonsovm44/gendox/master/installer.ps1 | iex

Interactive chat

Good for onboarding.

gendox query "how does auth work in this project?"
# or you can open a chat session
gendox query --chat
# or open a project-specific session:
gendox query --chat --project .  # reads .docgen/docs and project context

This launches a terminal chat where the AI answers questions about the documented codebase and links back to generated pages.

Core concepts (short)

• Docs-as-Code: generated Markdown lives next to source, tracked in .docgen/.
• Incremental builds: file-level hashing avoids re-generating unchanged docs.
• Context-aware (RAG): dependency analysis provides the LLM with the right context for complex APIs.

Docfile — example (store the blueprint for Gendox)

Track:
    main.cpp
    src/
Ignore:
    src/secret.cpp

Style:
    -dont use emojis
    -be professional
    -be concise

Config & connection

Run gendox config to manage the backend and model:

gendox config mode local    # default (Ollama)
gendox config model qwen2.5-coder:7b
gendox config mode cloud
gendox config protocol openai
gendox config key <YOUR_API_KEY>
gendox config check           # verify connection

Configuration keys quick reference

Key	Description	Example
mode	Backend mode	local / cloud
protocol	API format	openai / google / simple
key	Cloud API key	sk-...
model	Model identifier	gpt-4o, qwen2.5-coder:7b

FAQ (short)

Q: Is gendox rewriting my code?
A: No. gendox reads source and generates Markdown docs. It never mutates your source files unless you explicitly run commands that write generated artifacts.

Q: Is it deterministic?
A: gendox hashes context and caches outputs. Determinism depends on model behavior. For full reproducibility, pin model + prompt versions and use the cache/artifact files. gendox reuses docs so they are not rewritten from scratch everytime a tracked file is modified. It is incremental.

Contributing & support

• Please open issues for bugs/feature requests and PRs for examples and docs. See CONTRIBUTING.md and SECURITY.md for reporting sensitive issues.

License

• MIT — see LICENSE file.