Geometry NodeSet — AgentCanvas

Deliberately cross-method geometry utilities. Provides the distance-measurement canvas node; the former frontier-scoring mock was split off to the quarantined mock_frontier nodeset (other/mock_frontier.py) in the 2026-06-11 role-directory migration (ADR-platform-008).

1. Overview

The Geometry NodeSet (geometry, at workspace/nodesets/common/geometry.py) provides lightweight utility tools that complement domain-specific nodesets. The companion Mock Frontier NodeSet (mock_frontier) is documented here too until it is promoted or deleted:

Node	Purpose	Category
Measure Distance	Compute Euclidean distance between 3D points	Spatial Math
Score Frontier	Rank exploration frontiers by priority	Exploration

These nodes are self-contained — they do not require environment initialization or external service dependencies.

2. Canvas Nodes

Node Type	Display Name	Input Ports	Output Ports	Description
`geometry__measure_distance`	Measure Distance	`point_a` (TEXT), `point_b` (TEXT)	`result` (TEXT)	Calculate Euclidean distance between two 3D points. Inputs are JSON-formatted coordinate arrays `[x, y, z]`. Returns JSON result.
`mock_frontier__score_frontier`	[Mock] Score Frontier	`frontiers` (TEXT)	`result` (TEXT)	Score a list of frontiers by exploration priority. Mock implementation — returns random placeholder scores. Sorts results by priority.

UI Colors: - Both nodes use emerald color

Measure Distance

Computes the Euclidean distance between two 3D points.

Input format:

{
  "point_a": "[x1, y1, z1]",
  "point_b": "[x2, y2, z2]"
}

Points can be supplied as: - JSON string: "[1.0, 2.0, 3.0]" - Python list (if wired from another node): [1.0, 2.0, 3.0]

Output:

{
  "result": "{\"distance\": 5.1962}"
}

Distance is rounded to 4 decimal places.

Example:

point_a = [0, 0, 0]
point_b = [3, 4, 0]
distance = sqrt(3² + 4²) = 5.0

Score Frontier

Ranks exploration frontiers by priority. This is a mock implementation for development and testing — scores are randomly assigned rather than based on real frontier evaluation.

Input format:

{
  "frontiers": "[{\"id\": 1, \"center\": [0, 0, 0]}, {\"id\": 2, \"center\": [1, 1, 0]}]"
}

Expects a JSON string containing a list of frontier objects (e.g., from a frontier detection node).

Output:

{
  "result": "{\"scored_frontiers\": [{\"id\": 1, \"center\": [...], \"score\": 0.85, \"priority\": \"high\"}, ...], \"total\": 2}"
}

Each frontier receives: - score: Random value 0.1–1.0 (rounded to 3 decimals) - priority: Derived from score: - score > 0.7 → "high" - score > 0.4 → "medium" - score ≤ 0.4 → "low"

Results are sorted by score in descending order (highest score first).

Note on mock status: This node is intended as a placeholder for future frontier evaluation methods (visual saliency, distance, semantic importance, etc.). For production use, replace with a learned or heuristic scoring function.

3. Usage

Loading the NodeSet

Send a POST request to the backend:

POST /api/components/nodesets/geometry/load

Both nodes become available in the canvas UI under their respective categories (Spatial Math, Exploration).

Wiring Patterns

[Agent State] → extract position A
            → [Measure Distance] ← position B from target
            → [LLM: Distance-aware decision]

Pattern 2: Frontier Exploration (Development)

[Frontier Detector] → frontiers JSON
                   → [Score Frontier] (mock)
                   → [LLM: Select next frontier]

Caution: The frontier scorer is a mock. Do not rely on its output for production exploration strategies.

Pattern 3: Combining Both

[Frontier List] → [Score Frontier] → [Select top-ranked frontier]
                                   → [Measure Distance] to agent position
                                   → [Navigation Decision]

Error Handling

Invalid JSON: The nodes accept both string and list inputs; malformed JSON will raise an exception.
Missing keys: Frontier scorer gracefully handles missing fields by copying them unchanged.
Empty frontiers: Empty input returns {\"scored_frontiers\": [], \"total\": 0}.