debug-get-local-grid_skill

Overview

This skill provides a debug endpoint that returns the raw local_grid structure from the agent's world_state. It is intended for testing and inspecting the coordinate system, grid layout, and per-cell metadata. Use it to verify that the agent's local mapping and coordinate transforms are correct.

How this skill works

When invoked with a direct debug call, the skill reads the world_state and extracts the local_grid object without modification. It returns a uniform JSON with a human-readable summary in value and the unaltered local_grid dict in data. The local_grid contains center coordinates, integer radius, and a dictionary of cells keyed by "x,y,z".

When to use it

• Validating coordinate transforms and agent pose reflected in the local grid
• Inspecting cell-level data (occupancy, flags, or custom metadata) during development
• Debugging mismatches between sensor data and internal map representation
• Unit testing pathfinding, navigation, or grid generation routines
• Quickly exporting the internal grid for logging or offline analysis

Best practices

• Invoke only in development or controlled test environments; output is raw and may expose internal structure
• Compare the returned center x,y,z,yaw with ground truth to detect transform errors
• Use small radii or filtered queries when inspecting large grids to reduce noise
• Log or snapshot outputs for repeatable test cases and regression checks
• Pair with visualizers or mapping tools to interpret the cell key layout ("x,y,z")

Example use cases

• Call the skill after a localization update to confirm the local_grid center moved as expected
• Run before and after a sensor fusion change to ensure cell values remain consistent
• Extract raw grid to reproduce a navigation failure in a unit test
• Validate that a new grid-generation routine populates cell keys and metadata correctly
• Use in CI to assert the presence and basic shape of the local_grid structure