neurocore.runtime.blueprint

Blueprint parser and validator for NeuroCore.

A blueprint is a YAML file that defines a FlowEngine flow using skill names (not Python class paths). The runtime resolves skill names to classes via the SkillRegistry.

Blueprint format:

name: "my-flow"
version: "1.0"
description: "Optional description"
components:
  - name: memory
    type: neuroweave          # Skill name from registry
    config:
      llm_model: "claude-haiku-4-5-20251001"
  - name: search
    type: web-search
    config:
      max_results: 10
flow:
  type: sequential
  steps:
    - component: memory
    - component: search

Classes

BlueprintComponent	A component definition in a blueprint.
FlowStep	A step in a sequential or conditional flow.
FlowGraph	Graph node definition.
FlowEdge	Graph edge definition.
FlowDefinition	Flow structure definition.
Blueprint	Complete blueprint model.

Functions

load_blueprint(→ Blueprint)	Load and parse a blueprint YAML file.
validate_blueprint(→ list[str])	Validate that all skill references in a blueprint can be resolved.

Module Contents

neurocore.runtime.blueprint.load_blueprint(path: pathlib.Path) → Blueprint

Load and parse a blueprint YAML file.

Parameters:

path – Path to the blueprint YAML file.

Returns:

Parsed Blueprint model.

Raises:

BlueprintError – If the file cannot be read, parsed, or validated.

neurocore.runtime.blueprint.validate_blueprint(blueprint: Blueprint, registry: neurocore.skills.registry.SkillRegistry) → list[str]

Validate that all skill references in a blueprint can be resolved.

Checks that every components[].type matches a registered skill.

Parameters:

• blueprint – Parsed blueprint.

• registry – SkillRegistry with discovered skills.

Returns:

List of validation error messages (empty if valid).