> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/nemo/relay/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/nemo/relay/_mcp/server.

# About

Use this section when you want the `nemo-relay` binary to observe local coding
agent sessions through hooks, a passthrough LLM gateway, and NVIDIA NeMo Relay
observability exporters.

Install the NeMo Relay CLI with the curl-based installer described in
[Installation](/getting-started/installation). The installer provides the
`nemo-relay` binary without requiring a Rust toolchain. The CLI can run supported
coding agents through a managed local gateway, forward agent hook payloads into
NeMo Relay lifecycle events, route OpenAI-compatible or Anthropic-compatible
model traffic through the gateway, and diagnose local configuration. The
`nemo-relay-cli` Cargo package remains available for source-based installation
and unsupported platforms.

Persistent integrations use one lifecycle path:

`Codex / Claude Code` → `nemo-relay mcp` → `shared Relay gateway`

The host-specific layer installs and emits canonical lifecycle hooks. The MCP
process is host-neutral and keeps the shared user-level gateway available.

## Start Here When

Use these guides when you need to:

* Observe Claude Code, Codex, or pi sessions locally.
* Configure coding-agent hooks for NeMo Relay lifecycle events.
* Route model-provider traffic through the local NeMo Relay gateway.
* Install persistent Claude Code, Codex, or pi integrations without
  wrapping the agent command.
* Gate pi tool calls on a Relay policy from inside a pi extension.
* Validate and install model pricing catalog sources for local cost estimates.
* Export local sessions to Agent Trajectory Interchange Format (ATIF), Agent
  Trajectory Observability Format (ATOF) JSONL, OpenTelemetry, or
  OpenInference.
* Diagnose hook loading, gateway routing, and exporter output.

If you are instrumenting an application or framework directly, use
[Instrument Applications](/instrument-applications/about) or
[Integrate into Frameworks](/integrate-into-frameworks/about) instead.

## Agent Harness Support

The NeMo Relay CLI is a supported surface for local coding-agent observability
and hook-backed security. Optimization remains partial and host-dependent
because each agent exposes different hooks, provider routes, and execution
controls.

| Agent       | Observability | Security | Optimization | Notes                                                                                                                                                                                                                                                                                                                                             |
| ----------- | ------------- | -------- | ------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Claude Code | Yes           | Yes      | Partial      | Pre-tool hook responses are supported. LLM optimization uses gateway-routed traffic; full coverage depends on loaded Claude Code hooks.                                                                                                                                                                                                           |
| Codex       | Yes           | Yes      | Partial      | Persistent install verifies all 10 hooks in the supported schema. Each `Stop` finalizes a turn snapshot because the plugin schema does not expose `SessionEnd`.                                                                                                                                                                                   |
| pi          | Partial       | Yes      | No           | Proof of concept. Tool and turn activity are captured through a Relay-authored pi extension, and both tool calls and model calls can be blocked. Model traffic reaches the gateway only when the gateway forwards to the endpoint every model of the selected provider would otherwise call, because pi's provider registration is provider-wide. |

For minimum agent versions, platform support, and current limitations, refer to
the [Support Matrix](/reference/support-matrix).

## Guides

Use these guide links to move from CLI setup into agent-specific instructions.

* [Basic Usage](/nemo-relay-cli/basic-usage) explains gateway routes, transparent runs,
  shared configuration, hook forwarding, and runtime mapping.
* [Managed Daemon](/nemo-relay-cli/daemon) covers the brokered multi-user
  daemon, managed hook and MCP commands, worker networking, immutable settings,
  and lossless streaming transport.
* [Coding Agent Installation](/nemo-relay-cli/plugin-installation) covers
  persistent Claude Code, Codex, and pi setup.
* [Claude Code](/nemo-relay-cli/claude-code) covers transparent Claude Code
  runs, Anthropic gateway routing, ATIF verification, and unsupported Claude
  application modes.
* [Codex](/nemo-relay-cli/codex) covers transparent Codex CLI runs, local
  GUI/app caveats, model provider routing, and remote-task limits.
* [pi](/nemo-relay-cli/pi) covers the extension-based pi integration, tool
  gating, agent-run attribution, and the model-redirection gap.

Start with [Basic Usage](/nemo-relay-cli/basic-usage), then use the guide for the coding
agent that you want to observe.

Hermes Agent includes NeMo Relay natively and does not use this CLI. No
separate observability plugin or Relay setup is required. Hermes Agent
understands NeMo Relay plugin configurations.