Getting Started

Quick Start

View as Markdown

Choose a Quick Start path based on the boundary that you want Relay to observe or control:

  • Use the CLI path when a local coding-agent harness owns the invocation.
  • Use a binding quick start when your application owns the tool or LLM call.
  • Use a supported integration guide when a framework or agent harness owns the boundary.

If you still need the shared runtime vocabulary for scopes, middleware, events, or plugins, refer to the Agent Runtime Primer or Concepts first.

Using an AI Coding Assistant?

Install NVIDIA agent skills to give compatible coding agents product-specific Relay workflows and references. After the NeMo Relay skills are published in nvidia/skills, install the full Relay skill set without cloning the catalog:

$npx skills add nvidia/skills \
>  --skill nemo-relay-install \
>  --skill nemo-relay-get-started \
>  --skill nemo-relay-instrument-calls \
>  --skill nemo-relay-instrument-context-isolation \
>  --skill nemo-relay-instrument-typed-wrappers \
>  --skill nemo-relay-plugin-observability \
>  --skill nemo-relay-plugin-adaptive-tuning \
>  --skill nemo-relay-plugin-build \
>  --skill nemo-relay-migrate-from-flow \
>  --skill nemo-relay-debug-runtime-integration \
>  --agent codex

Use --agent claude-code, --agent cursor, or another supported agent target when that is where you want the skill installed. To inspect the current NVIDIA skill catalog first, run npx skills add nvidia/skills --list.

Examples by Integration Layer

Use this map when you know what Relay boundary you want to verify, but you do not yet know which guide owns the working path.

LayerUse WhenStart HereSuccess Check
CLI and gatewayA coding-agent harness owns invocation and provider routingCLI Basic UsageThe wrapped agent runs with Relay active and emits hook output, plus gateway-routed LLM lifecycle output when provider routing is active.
Persistent coding-agent installsYou want the maintained install path for Codex, Claude Code, or Hermes Agent instead of the transparent wrapperCoding Agent Installationnemo-relay doctor --plugin <host> confirms the installed integration is ready.
Direct Python or Node.js application APIsYour application owns the tool or LLM callbackPython Quick Start or Node.js Quick StartThe sample prints event lines plus tool and LLM results.
Direct Rust application APIsYour Rust application owns the tool or LLM callbackRust Quick StartThe sample prints scope, tool, and LLM lifecycle output plus the initialized mark event.
Plugin-managed runtime setupYou need process-level exporter or plugin behavior from plugins.tomlPlugin Configuration FilesThe selected plugin path activates and writes the expected output or behavior.
Managed middlewareYou want policy, redaction, routing, or execution wrapping around managed callsAdd MiddlewareOne allowed request succeeds, one rejected request stops before execution, and observed payloads match the policy.
Framework integrationsA framework such as LangChain, LangGraph, or Deep Agents owns callbacks or schedulingSupported IntegrationsThe integration guide’s verify step confirms the expected framework-owned output.
OpenClaw plugin pathOpenClaw owns plugin setup and Relay observes the OpenClaw-managed boundaryOpenClawThe OpenClaw guide’s verify step confirms plugin setup and runtime output.
Manual or CI workflowsYou need explicit config files, deterministic commands, or non-interactive automationCLI Basic Usage and Plugin Configuration FilesThe explicit command uses the intended config files without interactive setup.

Local Coding-Agent Runs

Use the NVIDIA NeMo Relay CLI when you want to observe a local Codex, Claude Code, or Hermes Agent session without changing application code.

CLI Basic Usage

Run a coding-agent CLI through Relay, forward hooks, route model traffic through the local gateway, and export traces or trajectories.

Framework or Agent-Harness-Owned Code

Use a supported integration guide when LangChain, LangGraph, Deep Agents, or OpenClaw owns scheduling, callbacks, tool execution, provider payloads, or agent lifecycle hooks. If you are building or reviewing a framework integration that is not covered by a maintained guide, use the generic framework integration workflow instead.

Supported Integrations

Choose the maintained integration path and support level for LangChain, LangGraph, Deep Agents, or OpenClaw.

Integrate Into Frameworks

Choose the generic workflow for unsupported frameworks, provider adapters, or new public-API integrations.

Framework Integration Concepts

Ensure you understand what the framework owns, what Relay owns, and which runtime semantics each integration boundary preserves.

Application-Owned Code

Use a binding Quick Start when your application owns the LLM or tool call and you want to instrument that boundary directly.

Python Quick Start

Smallest Python workflow that emits scope, tool, and LLM events.

Node.js Quick Start

Smallest Node.js workflow that emits scope, tool, and LLM events.

Rust Quick Start

Smallest Rust workflow for direct runtime ownership that emits scope, tool, and LLM events.