The TestRelic Ask AI agent in your terminal — a Claude Code-style interactive CLI with a full MCP host, written in Rust.

The TestRelic CLI runs the same Ask AI agent the web app uses — in a Claude Code-style inline terminal UI, written in Rust. It streams markdown, shows live tool activity, renders artifacts (dashboards, reports, test plans) directly in your scrollback, and acts as a full MCP host for your local tools.

One binary, three names:

testrelic                                       # interactive UI
testrelic -p "why did the nightly regress?"     # print mode (pipes welcome)
tr                                              # same binary, shorter name
ask-ai                                          # same binary, mnemonic alias

Install in one line

Linux / macOS: curl -fsSL https://testrelic.ai/install.sh | sh Windows (PowerShell): irm https://testrelic.ai/install.ps1 | iex

See Installation for targets, versions, and environment variables.

What it does

Interactive agent chat against the TestRelic cloud — streamed markdown, live tool activity (⏺ tool(args) / ⎿ result), artifacts, and token usage, in an inline terminal UI that scrolls into native scrollback (no alt-screen).
Every artifact renders in the terminal — New Relic-grade dashboards (KPI billboards, sparklines, braille charts, heat-colored tables), reports, test plans, navigation paths, data tables, code, and session cards. Live generation shows a streaming progress line.
Tool confirmations — destructive tools (repo memory writes, alert CRUD) pause the stream for an Approve / Reject panel.
@ context and ! shell mode — attach repos, runs, branches, tests, environments, suites, or tags; run a local command whose output attaches to your next message.
Full MCP host — run local MCP servers (stdio or HTTP) whose tools the cloud agent can call. See MCP host.
Headless print mode — -p with --output-format text|json|stream-json for scripts and CI. See Artifacts & headless output.

Commands at a glance

Beyond the interactive agent, the CLI exposes commands for onboarding, IDE integration, and local config:

Command	What it does
`setup`	Deterministic, keys-only onboarding — `--api-key tr_live_`, `--mcp-token tr_mcp_`, `--no-install`. Wires up local config so the SDKs can upload.
`init`	Agentic onboarding that inspects your project — `--yes`, `--framework`, `--no-upload`.
`login` / `logout`	Authenticate or sign out. See Authentication.
`upload`	Import existing Allure/JUnit reports into TestRelic Cloud. See Upload reports.
`map`	Save a Project Map to your cloud repo memory.
`mcp add\|list\|remove`	Manage local MCP servers the agent can call. See MCP host.
`studio`	Launch the local studio — `--path`, `--stop`.
`results` / `run` / `complete`	IDE Test Explorer integration.
`usage`	Ask AI token usage against your org budget.
`artifacts`	List, show, and export artifacts. See Artifacts.
`config get\|set\|list\|path`	Read and write local configuration. See Configuration.
`doctor`	Auth, MCP, and terminal diagnostics.

See Modes & engines for the --mode and --engine axes that apply across the interactive agent.

Where to go next

Install

One-line install for Linux, macOS, and Windows, with version pinning and environment variables.

Quickstart

Authentication

Session login for full parity, or a personal access token for headless and CI use.

MCP host

Add local MCP servers so the cloud agent can call your tools during a turn.

How it relates to the rest of TestRelic

The CLI is a client of the same cloud Ask AI agent surfaced in the web app and over MCP. The SDK reporters send your test runs to the cloud; the CLI lets you query and act on that data from the terminal. Authentication mirrors @testrelic/mcp — the same tr_mcp_* token convention works for both.

Distribution

The CLI ships as prebuilt binaries for Linux, macOS, and Windows — there's nothing to compile. Install with the one-liner above; see Installation for supported targets and version pinning.

Next steps

→ Install the CLI

TestRelic CLI overview