Core Concepts
This section explains the main ideas that power Raiken.
At a glance
| Concept | Purpose | Stored in |
|---|---|---|
| Code graph | Structural context for prompts | .raiken/raiken.db |
| DOM capture | UI context and selectors | .raiken/ cache |
| Orchestrator | Routes the right tools | CLI process |
| Test generation | Produces test files | tests/ directory |
Code graph
Raiken builds a graph of your project to provide structural context. This helps the agent understand entry points, dependencies, and file layout.
Where it lives:
.raiken/raiken.db
DOM context
Raiken captures a semantic snapshot of the page to generate stable selectors. This is optimized for test authoring, not full page scraping.
See: DOM capture
Orchestrator
The orchestrator decides which tools to call:
- code graph for file context
- DOM capture for UI context
- test generation for output
This routing keeps the agent focused on the right context for each request.
Local state
Raiken stores local project state on disk so it can:
- avoid re-indexing on every run
- reuse prior context
- improve test generation accuracy over time
Test generation
Prompts are grounded by:
- code graph context
- DOM context
- user intent
The result is a test file (Playwright by default) with selectors derived from actual UI structure.