AI code documentation

Generates a security-oriented tutorial of a codebase: what the main abstractions are, how they relate, and what a reviewer should understand before reading the findings. Enable it with AI Code Documentation → Generate AI Code Documentation on the Static Analysis page.

How it works

The pipeline crawls the source files for the selected language and then runs a sequence of model passes:

1. 01
   Identify abstractions

   What are the significant concepts in this codebase?

2. 02
   Analyze relationships

   How do those abstractions depend on and call each other?

3. 03
   Order chapters

   What order should a newcomer read them in?

4. 04
   Write chapters

   One chapter per abstraction, written for a reader who has not seen the code.

5. 05
   Combine

   The chapters are assembled into a single tutorial.

Reading the output

The generated documentation is packaged as an archive and published so it can be read directly in the web interface. It is also available from the Report button in Scan History.

Cost

Documentation generation reads a large portion of the codebase, so it is one of the more expensive AI features per run. It is also the one you are least likely to need repeatedly — generate it once when you take on an unfamiliar codebase, rather than on every commit.