plato-tile-pipeline

One-call tile pipeline — validate, score, dedup, store, search, rank.

Part of the PLATO framework — deterministic AI knowledge management through tile-based architecture.

Installation

pip install plato-tile-pipeline

Usage

from plato_tile_pipeline import TilePipeline

pipeline = TilePipeline()

result = pipeline.process(
    tiles=[
        {"content": "Pythagorean triples snap to exact coordinates", "confidence": 0.9, "domain": "constraint-theory"},
        {"content": "Deadband Protocol: P0 rock, P1 channel, P2 optimize", "confidence": 0.85, "domain": "governance"},
        {"content": "Bad tile", "confidence": 0.1},  # rejected
        {"content": "Pythagorean triples snap to exact coordinates"},  # duplicate
    ],
    query="pythagorean",
)

print(result.processed)  # 4
print(result.accepted)   # 1 (bad rejected, dup removed)
print(result.tiles)      # [best matching tile]

The Pipeline

validate -> score -> dedup -> store -> search -> rank

Six PLATO packages behind one API call. The playset.

Zero external dependencies. Compatible with Python 3.8+.

GitHub