documentor

documentor

documentor is an easy-to-use, efficient, and powerful command-line application that uses the power of AI to review documentation and provide feedback on how to improve it, draft new documentation from scratch, and more.

It assists documentation writers in creating better documentation by automating parts of the writing and review process, allowing them to focus on the content and structure of the document itself.

Installation

From source

First, ensure that the following dependencies are installed:

• Go 1.24 or above.
• make.
• scdoc.

Optionally, you can install glow to render the Markdown output of the review command with more style.

Then, switch to the latest stable tag (v1.2.0), compile, and install:

git checkout v1.2.0
make
sudo make install

Usage

[!NOTE] If you're a Datadog employee, please follow the documentation in Confluence.

$ documentor --help
NAME:
   documentor - improve technical documentation with the power of AI

USAGE:
   documentor [global options] command [command options]

VERSION:
   1.2.0

COMMANDS:
   review, r    review technical documentation
   describe, d  describe an image and generate alt text
   draft, D     draft new documentation based on the provided notes

GLOBAL OPTIONS:
   --key value, -k value          the API key to use [$DOCUMENTOR_KEY]
   --provider value, -p value     the AI provider to use (default: "openai") [$DOCUMENTOR_PROVIDER]
   --endpoint value, -e value     the API endpoint to use (currently only used for the Datadog provider) [$DOCUMENTOR_ENDPOINT]
   --model value, -m value        the AI model to use (default: "gpt-4o") [$DOCUMENTOR_MODEL]
   --temperature value, -t value  the temperature to use for the model (default: 0.8) [$DOCUMENTOR_TEMPERATURE]
   --help, -h                     show help
   --version, -v                  print the version

Examples

1. Review a documentation file using the default provider, OpenAI:

documentor --key 'your-openai-api-key' review '/path/to/file.md'

2. Review a documentation file with the API key set in the environment:

export DOCUMENTOR_KEY='your-openai-api-key'
documentor review '/path/to/file.md'

3. Save the output to a file:

documentor review '/path/to/file.md' >> review.md

4. Format the default Markdown output with glow:

documentor review '/path/to/file.md' | glow

5. Describe an image:

documentor describe '/path/to/image.png'

6. Describe an image with context:

documentor describe --context 'This my cat, Mittens.' '/path/to/image.png'

7. Describe an image and generate a filename for the image:

documentor describe --filename '/path/to/image.png'

8. Draft a document based on your notes:

documentor draft '/path/to/notes/file.md'

9. Draft a document using Anthropic as the AI provider:

documentor --provider 'anthropic' --key 'your-anthropic-api-key' draft '/path/to/notes/file.md'

10. Review a document using a different LLM model:

documentor --model 'o3-mini' review '/path/to/file.md'

Refer to the documentor(1) manpage after installation for more information.

Contributing

Anyone can help make documentor better. Refer to the contribution guidelines for more information.

This project is released under the Apache-2.0 License.