Introduction

​

What Vidoc is

Vidoc is a code security review platform. It uses large language models (LLMs) to find security issues in source code. Then an agent examines the repository to confirm or reject each finding. An agent is an LLM program that does a task in steps. It reads files, searches the code, and then gives a result. Vidoc finds issues in four steps:

• Context. LLM agents read the code and the context of the repository, for example its entry points and trust boundaries.
• Validation. A validation agent examines the codebase for each high and critical finding. Vidoc shows only the findings that the agent confirms.
• Feedback. Vidoc keeps the reasons from triage decisions and pull request (PR) replies as learnings. The validation agent uses them in later scans.
• Output. Each finding has an explanation of the risk, the attack path, the reasoning of the validation, and a proposed fix.

For the full scan process, refer to How Vidoc works.

​

How you use Vidoc

Some features are available only on some source code platforms. For the full list, refer to Supported platforms. Vidoc is available as a cloud service (Vidoc Cloud) and as a self-hosted installation, including on-premises deployments. Open app.vidoc.dev or your custom deployment URL. For the differences, refer to Deployment options.

​

Next steps