What problem does it solve?
Applications that already own tool functions or LLM/provider call sites often lack lifecycle events, middleware, and guardrail integration. This Skill guides agents to wrap those existing call sites with NeMo Relay scopes and managed execution APIs without changing the original callable behavior.
Core Features & Use Cases
- Managed Execution Wrapping: Use language-specific managed APIs such as
tools.execute(...) in Python, tool_call_execute(...) in Rust, toolCallExecute(...) in Node.js, and tools.Execute(...) in Go to instrument existing tool and LLM calls.
- Runtime Semantics Guidance: Explains how conditional-execution guardrails, request intercepts, sanitize guardrails, and execution intercepts interact with the wrapped callback.
- Scope and Lifecycle Checklist: Helps place scope boundaries around agent or workflow boundaries and ensures every manual start call has a matching end or error path.
- Use Case: A Node.js support chatbot calls a
lookup_order tool and then GPT-4 for summarization; this Skill shows how to route both calls through NeMo Relay for observability and guardrail compliance while keeping business logic intact.
Quick Start
Ask your agent to wrap your existing tool functions and LLM provider calls with NeMo Relay managed execution APIs so they emit lifecycle events and pass through guardrails.
