Scenarios using Screph modules

Coordinates alone are insufficient for a scenario: it needs object identity, visual features, states and verification data.

A Windows screen or window, a still image or a recorded sequence.

The project stores elements, relations and annotations. OCR, matching and other computer-vision methods are optional; proposed results are reviewed separately.

A main project for Automation Runtime, the selected main.py, and a separate for_ai_agent context for Screph Code or another agent tool.

An OpenCV notebook alone does not link the region of interest, parameters, visual output and project changes.

An image, screenshot or validated set of videos and folders.

A region of interest is linked to a method, linear sequence or method graph. Source data is compared with the result, mask, contours, OCR and metrics.

Method parameters, an execution trace and separate result files. Changes to main-project elements and features are stored only after separate application.

A keyboard and pointer event log does not contain the reviewed visual object the user interacted with.

An action recording with frames before, during and after the action and with coordinate transformations.

An LLM proposes a method plan, computer-vision methods run locally, and proposed results are compared with the source frames and accepted, edited or rejected separately.

Review decisions, proposed targets, provenance data, a proposed project-graph change and enriched automation context.

General markup does not include domain-specific types for defects, measurement zones, areas of interest, changes and tracks.

An industrial image or region of interest, or a UAV frame or series with available temporal and spatial data.

A task for the selected mode is created, a configured method plan is executed, results are compared with the source data, and a supported element or region change is applied separately.

Main markup and separate mode files for further investigation.