[](https://github.com/NVIDIA/NeMo-Relay/blob/main/LICENSE)
[](https://github.com/NVIDIA/NeMo-Relay/)
[](https://github.com/NVIDIA/NeMo-Relay/releases)
[](https://app.codecov.io/gh/NVIDIA/NeMo-Relay)
[](https://pypi.org/project/nemo-relay/)
[](https://www.npmjs.com/package/nemo-relay-node)
[](https://www.npmjs.com/package/nemo-relay-wasm)
[](https://crates.io/crates/nemo-relay)
[](https://crates.io/crates/nemo-relay-adaptive)
[](https://crates.io/crates/nemo-relay-cli)
[](https://deepwiki.com/NVIDIA/NeMo-Relay)
# NeMo Relay
`nemo-relay-cli` installs the NeMo Relay CLI, the `nemo-relay` binary for local
coding-agent observability. It can configure supported coding-agent hooks, run
agents through an ephemeral gateway, and diagnose local agent and exporter
readiness.
The CLI is a Rust package in this repository, but most users should interact
with the installed `nemo-relay` command rather than link against the crate.
## Why Use It?
- 🧭 **Observe existing coding agents**: Run Claude Code, Codex, Cursor, or
Hermes Agent through a local NeMo Relay gateway without changing the agent
itself.
- 🛠️ **Configure hooks interactively**: Use the setup wizard to write project or
user config and install the hook files needed by supported agents.
- 📡 **Export local sessions**: Write ATIF trajectory files, ATOF event JSONL
streams, or OpenInference spans from one shared config model.
- 🩺 **Diagnose the machine**: Check config layers, agent binaries, hook status,
observability outputs, and shell completions with `nemo-relay doctor`.
## What You Get
- ✅ **`nemo-relay` binary**: The executable installed by the `nemo-relay-cli`
Cargo package.
- ✅ **First-run setup**: Bare `nemo-relay` launches setup when no config exists,
then runs doctor once config is present.
- ✅ **Agent shortcuts**: `nemo-relay claude`, `nemo-relay codex`,
`nemo-relay cursor`, and `nemo-relay hermes` start observed agent runs.
- ✅ **Config-driven launch**: `nemo-relay run` resolves config, environment, and
CLI overrides for deterministic non-interactive use.
- ✅ **Hook forwarding server**: A local gateway accepts agent hook events and
provider-shaped OpenAI or Anthropic requests.
## Installation
Install the CLI:
```bash
cargo install nemo-relay-cli
```
That command installs the binary as:
```bash
nemo-relay --version
```
## Getting Started
Run the first-time setup wizard:
```bash
nemo-relay
```
After setup, inspect local readiness:
```bash
nemo-relay doctor
```
Run a supported agent through the gateway:
```bash
nemo-relay codex
nemo-relay claude -- "summarize this repository"
```
Use `run --dry-run` to inspect resolved config without spawning the agent:
```bash
nemo-relay run --agent codex --dry-run
```
## Configuration
Project config lives at `./.nemo-relay/config.toml`; user config lives at
`~/.config/nemo-relay/config.toml` or `$XDG_CONFIG_HOME/nemo-relay/config.toml`.
The project layer overrides system config, and the user layer overrides the
project layer.
General options are configured through the top-level config. Edit the config with:
```bash
nemo-relay config
```
Observability exporters are configured through the plugin config. Edit the user
plugin config with:
```bash
nemo-relay plugins edit
```
The canonical plugin file is `plugins.toml`; user config lives at
`~/.config/nemo-relay/plugins.toml` or
`$XDG_CONFIG_HOME/nemo-relay/plugins.toml`. Project config lives at
`.nemo-relay/plugins.toml`.
Minimal ATIF example:
```toml
version = 1
[[components]]
kind = "observability"
enabled = true
[components.config.atif]
enabled = true
output_directory = "./atif"
```
## Documentation
NeMo Relay Documentation: https://nvidia.github.io/NeMo-Relay/