The following standardized process needs to be followed to use Engineering's document automation features:
- Code base access: Associate a GitHub/GitLab repository in the platform dashboard and the system will scan the code structure. It is recommended to choose a mature project that contains complete comments (at least 80% comment coverage), so that the platform can better recognize the intent of the code.
- Document Type Selection: The platform supports multiple document outputs:
- API documentation: automatic parsing of function/method signatures, parameter descriptions and return values
- System Manual: Generate overall architectural descriptions based on module annotations
- Deployment Guide
: Identify Dockerfile or CI/CD configurations to generate environment dependency descriptions
- Generation and Calibration: After clicking on "Generate Document", the system will:
- Extracting docstring and special tags (e.g. @param) from code
- Analyzing Call Relationships to Supplement Context
- Generate Markdown/HTML first drafts
It is recommended to manually proofread key sections, especially modules with complex business logic.
- continuous synchronizationEnable "auto update", when the relevant code changes, the document will be submitted for update via PR, and the team can set up review rules to ensure the quality of the document.
Best practice: Use standardized comment formats (e.g. JavaDoc, Python Docstring) in your code, and the platform can recognize Swagger/OpenAPI specifications with an accuracy of 95% or more. For legacy systems, it is recommended to run the platform's "Document Health Scan" function to locate critical areas of missing comments.
This answer comes from the articleEngineering: GitHub's automated code review, documentation and team reporting platformThe
