SlipboxKnowledgeRouting¶

Slipbox knowledge routing step that hosts the DKS knowledge+ruleset corpus and runs compile→index→route internally, emitting a compiled prompt ruleset plus per-record routed rule names and routing confidence for downstream Bedrock processing


SageMaker step type	`Processing`
Node type	internal (consumes upstream, produces downstream)
Container entry point	`slipbox_knowledge_routing.py`
Interface file	`steps/interfaces/slipbox_knowledge_routing.step.yaml`

Compute¶


Compute kind	`framework`
SDK class	`PyTorch` (SageMaker DLC via `image_uris.retrieve`)

Functionality¶

Slipbox knowledge routing script that COMPILES the DKS rule_.md corpus into an in-memory prompt ruleset (prompts.json + tool schema), INDEXES the pattern_/behavior_* corpus with an offline SentenceTransformer encoder into an in-memory routing index, and ROUTES each input record via cosine similarity + activation top-k to a set of routed rule names with a routing_confidence score. An internal consistency gate asserts the index linked_rules are a subset of the compiled rule_names. Emits the prompt_ruleset and the routed_records for downstream BedrockProcessing.

Inputs (dependencies)¶

Input	Type	Required	Compatible producers
`records`	`processing_output`	yes	TabularPreprocessing, CradleDataLoading, ProcessingStep
`knowledge_corpus`	`custom_property`	no	DummyDataLoading
`embedding_model`	`model_artifacts`	no	PyTorchModel, TokenizerTraining

Outputs¶

Output	Type
`prompt_ruleset`	`processing_output`
`routed_records`	`processing_output`

Consumers (downstream steps)¶

Steps that declare this step as a compatible input source:

Framework requirements¶

Package	Version
`boto3`	`>=1.34`
`pandas`	`>=2.0`
`pyarrow`	`>=15.0`
`pyyaml`	`>=6.0`
`sentence-transformers`	`>=3.0`
`numpy`	`>=1.26`

← Back to the Step Catalog