DOKIMA

Documentation

Set up Dokima and understand the review.

Find installation, activation, workspace setup, review, findings, command and data-handling guidance.

Documentation guides

  1. Setup for coding agents

    A step-by-step setup checklist, Linux hosting and private report publishing.

  2. Agent findings API

    Create a read-only key and let an agent review hosted evidence and fix recommendations.

  3. Getting started

    Linux and macOS installation, activation, runner setup and your first review.

  4. Working with findings

    Understand report fields, verification and developer annotations.

  5. Sharing status and results

    Private team publishing or local HTML reports on your internal server.

  6. Command guide

    Choose between the dashboard, steps, stages, sweeps and repair work.

  7. Data handling

    Know what Dokima stores locally and what depends on your runner.

How Dokima works

Three steps to a continuous security review.

  • 01Point Dokima at a local clone of your repository. Review state and reports are stored on your computer.
  • 02Configure Claude, Codex or OpenCode. Use OpenCode for other providers and local models. Model usage or hosting costs are separate; your model setup determines where source code is processed.
  • 03Run Dokima and let it work. It hunts for vulnerabilities across your codebase and keeps reviewing as your code changes.

Before you begin

Dokima brings the workflow. You bring the runner.

Dokima supports Claude and Codex directly, and other models through OpenCode, including Gemini, open models and privately hosted local models. Configure your provider and default model before running a review. Model compatibility and review quality vary. See the runner setup guide.

Read about data handling