CodeGuard AI

A Java team that wants to change how AI reviews its pull requests could use CodeGuard AI as a starting point. The Gumroad listing sells the Spring Boot application source, rather than access to a hosted service. It describes fetching pull-request changes, sending them to a chosen model and turning the response into structured findings that can be posted to GitHub.

Manual dashboard reviews and GitHub webhook triggers are both listed, alongside review history and Markdown or PDF exports. The practical appeal is being able to change the review logic and interface. It also means taking responsibility for hosting, credentials and maintenance. ProjectRadar inspected the sales page and its illustrations without buying or running the package.

The listed requirements include Java 17 or later, Maven 3.8 or later, a GitHub personal access token and access to OpenAI, Gemini or Ollama. The seller says provider selection is configurable. With OpenAI or Gemini, review context is processed by that provider; the Ollama option is described as using a local model. Self-hosting the application alone does not keep cloud-model requests on the same machine.

The dashboard includes quality, risk and confidence scores. Those are signals to investigate, not proof that a change is correct or secure. The listing does not establish detection accuracy, and this note is not a security assessment of the application or the code it reviews.

The public price starts at US$15 for a one-time source-code purchase. The seller lists installation, API and architecture documentation, a database script and a licence among the included files. Their contents and licence permissions were not available for this public-page check. OpenAI or Gemini usage is billed separately, and running a local model still requires suitable computing resources.