Overview

The verification layer for the agentic coding era — TestSprite CLI puts the full testing platform in your coding agent’s hands.

What is the TestSprite CLI?

TestSprite tests frontend and backend software in the cloud, against the live product, not mocks. The CLI puts that platform in your coding agent’s hands: structured output it can parse, exit codes it can branch on, and one failure bundle it can act on.

The CLI shares the same backend, projects, and tests as the Web Portal and the MCP Server. A test you create from the terminal shows up in the dashboard, and vice versa. The --output json shape and exit codes are a stable contract for agents and CI.

Why a CLI for coding agents?

Without it, you run the app, screenshot the result, and paste it back to the agent. With it, the agent triggers a real cloud run and reads the verdict itself.

What broke comes back as one self-consistent failure bundle, all anchored to the same snapshot; the CLI never stitches data from two runs:

In the bundle	What it contains
Failing step	The exact step that broke, plus its neighbors for context.
DOM snapshots	The DOM at the point of failure — your agent can read it without a vision model.
Fix guidance	A root-cause hypothesis and a recommended fix target.

Prepend --dry-run to any command to see a canned sample in the real response shape, with no API key.

The verification loop

1. Install and onboard

   npm install -g @testsprite/testsprite-cli
   testsprite setup --agent AGENT
   

   setup stores your API key and installs the verification skill into your agent’s project, so the agent knows when and how to run tests. See Installation.

2. Create and run a test

   Describe the behavior as a plan file, then one command creates the test, runs it in the cloud against your live app, and blocks until a verdict. Exit 0 means passed, 1 means failed.

   testsprite test create \
     --project proj_8f0f6 --type frontend \
     --plan-from ./checkout-flow.plan.json \
     --run --wait --output json
   
3. On failure, read the bundle

   mkdir -p .testsprite
   testsprite test failure get test_3a9f21c7 --out ./.testsprite/failure
   
4. Fix and rerun

   testsprite test rerun test_3a9f21c7 --wait
   

   Every pass is banked into the durable suite, so the next change reruns rather than recreates. See Rerun & Auto-Heal.

Who It’s For

Audience	What they get
AI coding agents Primary	Claude Code, Codex, Cursor, Cline, Antigravity, Windsurf, Copilot, and more — the agent drives the loop without a human in the middle.
Developers at a terminal	The same surface, with predictable commands, consistent flags, and machine-readable output.
CI pipelines	Stable --output json contract, stable exit codes, and non-interactive --from-env auth.

One Platform, Three Surfaces

All three share the same tests, projects, and runs:

Web Portal

For humans: visual project setup, PRD uploads, dashboards, scheduling, billing and member management. Not scriptable.

Go to Web Portal docs

MCP Server

Lives in your IDE; the agent calls it as a tool during a session, including for local frontend testing. Session-scoped, not scriptable, not CI-compatible.

Go to MCP Server docs

CLI

The scriptable, agent-safe, CI-compatible surface: structured JSON, stable exit codes, non-interactive operation and durable run artifacts. Billing management uses the Portal. To test an app on your own machine, see Environments.

Capability	Web Portal	MCP Server	CLI
Create and run tests
Visual dashboards		—	—
Scheduling		—
localhost targets via tunnel	—		*
Agent-driven	—
Scriptable / CI	—	—
Structured failure bundles	—
Billing & org management		—	—

* Frontend runs against loopback addresses; see Environments.

Where to Go Next

Installation

Install the CLI and sign in in under 2 minutes

Quickstart

Create a test, run it, and read what broke — end to end

Key Terms

Projects, tests, runs, and the concepts behind them

Command Reference

Every command, flag, and exit code