Runnable examples

neuron-js includes first-party examples that can be executed from a clean checkout. Each example keeps the rule definition, input context, expected output, and runner separate so developers and coding agents can inspect the full contract before changing code.

Run all examples

From the repository root:

yarn examples

The command builds the package, then runs:

• examples/pricing-rules/run.ts
• examples/eligibility-check/run.ts
• examples/workflow-routing/run.ts
• examples/generic-decision-runtime/run.ts
• examples/n8n-code-node/run.ts
• examples/langgraph-decision-node/run.ts

Each runner exits with a non-zero status if the actual output differs from expected-output.json.

Example catalog

Pricing rules

Path: examples/pricing-rules/

Demonstrates a cart-pricing decision stored as JSON. The script checks a cart subtotal and applies a VIP discount through a registered action.

Files:

• rules.json — serializable script.
• input.json — execution context.
• expected-output.json — verified output summary.
• run.ts — executable TypeScript runner.

Eligibility check

Path: examples/eligibility-check/

Demonstrates an approval decision. The script checks an applicant score and writes an approved eligibility decision into context.

Files:

• rules.json — serializable script.
• input.json — execution context.
• expected-output.json — verified output summary.
• run.ts — executable TypeScript runner.

Workflow routing

Path: examples/workflow-routing/

Demonstrates deterministic routing for automation workflows. The script checks ticket priority and assigns an escalation route with an SLA.

Files:

• rules.json — serializable script.
• input.json — execution context.
• expected-output.json — verified output summary.
• run.ts — executable TypeScript runner.

Generic decision runtime

Path: examples/generic-decision-runtime/

Demonstrates the pure decision runtime with a caller-defined context schema, outcome schema, local effect-free components, receipt identity, and no vertical application dependency.

Architecture guide: docs/concepts/agentic-decision-architecture.md shows the host/LLM boundary: LLM extraction is advisory, the host resolves the canonical Decision Context, Neuron-JS evaluates an approved DecisionDefinition, and host-only side-effect routing consumes receipt-backed results.

Files:

• definition.json — decision definition with schemas, manifest, and script.
• context.json — caller-supplied JSON context snapshot.
• expected-output.json — verified status, outcome, diagnostics, and receipt identity.
• run.ts — executable TypeScript runner using evaluateDecision.

n8n Code node

Path: examples/n8n-code-node/

Demonstrates deterministic workflow routing for n8n. The script checks support-ticket risk signals and returns a human-escalation route with an SLA.

Files:

• rules.json — serializable script.
• input.json — execution context.
• expected-output.json — verified output summary and explanation metadata.
• run.ts — executable TypeScript runner.

LangGraph decision node

Path: examples/langgraph-decision-node/

Demonstrates LLM extraction/classification followed by deterministic Neuron-JS decisioning. The script routes a high-risk refund request to human review.

Files:

• rules.json — serializable script.
• input.json — execution context.
• expected-output.json — verified output summary and explanation metadata.
• run.ts — executable TypeScript runner.

Why this structure

The examples are intentionally data-first:

• JSON rule files can be stored, versioned, reviewed, or generated.
• Input files make scenario testing repeatable.
• Expected outputs define a clear contract for humans and AI coding agents.
• TypeScript runners show the minimal registry setup required by each scenario.