mc-local-grid-visualize_skill

Overview

This skill generates an interactive 3D HTML visualization of the session-local rolling occupancy grid and opens it in the default web browser. It renders voxels as colored cubes and includes controls for rotate, pan, and zoom so you can inspect what the agent perceives around itself. The tool produces a self-contained HTML snapshot that you can refresh after re-running to view updates.

How this skill works

The skill reads the current local occupancy grid from the agent runtime (executor.get_world_state("local_grid")). It converts each occupied cell into a colored voxel and writes a standalone HTML file using Three.js from a CDN. The generated file is automatically opened in the default browser and shows an agent center marker, a radius boundary, and interactive hover tooltips for block name and coordinates.

When to use it

• After performing observations or actions (mc-observe, mc-dig, mc-place) to inspect the updated grid.
• When debugging navigation, collision recovery, or path-planning decisions based on nearby blocks.
• To verify what the agent currently 'sees' and confirm sensor/voxel mapping correctness.
• When you need a quick visual snapshot of the agent's local environment without running long simulations.

Best practices

• Run the tool immediately after modifying the world state and refresh the browser to compare snapshots.
• Specify a custom output_file when saving multiple snapshots to avoid overwriting.
• Use hover tooltips and zoom to identify specific blocks and coordinates for targeted debugging.
• Keep the browser open while iterating; re-run the skill and refresh to view new snapshots.

Example use cases

• Visualize the occupancy grid after a failed pathfinding attempt to see unexpected obstacles.
• Confirm that newly placed or dug blocks are registered correctly in the local grid.
• Share a static HTML snapshot with teammates for offline inspection and troubleshooting.