What's Included

What the TestSprite CLI can do today, what’s on the roadmap, and what stays in the Web Portal or MCP.

The TestSprite CLI is production-ready for the workflows listed below, with a stable --output json contract that agents and CI can rely on today. Changes are strictly additive: new fields may appear, but existing fields, exit codes, and command shapes will not change.

Stability promise: the JSON output contract and exit codes are stable. New commands or flags may be added, but nothing currently documented will be removed or renamed without a major version bump.

What’s included

Auth

Capability	Command
Configure an API key and install the agent skills (with --from-env for CI)	setup
Verify the active profile	auth status
Remove credentials	auth remove

Diagnostics

Capability	Command
One-shot environment check — CLI version, Node runtime, profile, endpoint, credentials, live connectivity, and verify-skill install status	doctor

Projects

Capability	Command
List, get, create, update, and delete projects (frontend and backend types; delete cascades to everything under the project and requires --confirm)	—
Set the static backend credential injected into every backend test (free tier)	project credential
Configure recurring-token auto-refresh login for backend tests (Pro)	project auto-auth

Tests

Capability	Command
List and get tests	—
Create a single test	test create
Bulk-create frontend tests	test create-batch, test create-batch --plan-from-dir
Update test metadata	—
Delete a single test or bulk-delete	test delete, test delete-batch

Authoring

• Frontend: create from an agent-supplied plan file (--plan-from) with full planSteps[]; replace plan steps (test plan put)
• Backend: create from a test code file (--code-file, executed as Python — requests + pytest-style); declare captured variables (--produces/--needs) and teardown tests (--category teardown) for wave-ordered dependency runs; replace code with ETag-guarded optimistic concurrency (test code put --language python)
• Read the generated test code (test code get)
• Scaffold a schema-correct starter plan or backend code skeleton, entirely offline (test scaffold)
• Lint plan/steps files offline with the same validators create runs, before spending a network round-trip (test lint)

Running

Capability	Command
Trigger a single test run	test run
All tests in the project, in wave order	test run --all
Block until terminal	--wait, --timeout
Resume a detached run — one run ID, or several at once under one shared timeout	test wait <run-id...>
Create and run in a single command	test create --run --wait
Export a JUnit XML report after a batch --wait run (also on batch test rerun --wait)	--report junit --report-file <path>

Rerun & auto-heal

Capability	Command
Replay a frontend test verbatim (no credit charge) or re-run a backend dependency closure	test rerun
Rerun all tests matching a filter	test rerun --all
Auto-heal on by default for frontend reruns — uses a small amount of credit only when AI healing actually engages; opt out	--no-auto-heal
Replay a test repeatedly with auto-heal off and report a stability score	test flaky

Reading results

Capability	Command
Cumulative step log, scoped to a run with --run-id	test steps
Latest result, with inline analysis via --include-analysis	test result
Run history list, filterable by source and time range	test result --history
One-screen failure triage	test failure summary
Full failure bundle for latest failing run	test failure get
Run-scoped, immutable failure bundle — unaffected by concurrent runs	test artifact get <run-id>
Compare two runs — verdict, failureKind, failed step, code-version drift	test diff <run-a> <run-b>

Onboarding

Capability	Command
One command to configure credentials and install the agent skills, with a unified summary	testsprite setup

Coding-agent skill

Capability	Command
Write the TestSprite skill files — verification-loop (testsprite-verify) and first-run onboarding (testsprite-onboard) — for Claude Code (GA), or Cursor, Cline, Antigravity, Kiro, Windsurf, Copilot, and Codex (all experimental except Claude)	agent install
See all supported targets, modes, and landing paths	agent list
Check installed skill files against this CLI version (ok/stale/modified/unmarked/absent/corrupt)	agent status

Global capabilities

Capability	Command
Try any command offline without an API key	--dry-run
Stable, parseable contract for agents and CI	--output json
Named profiles and environment variables for flexible credential management	—
On run completion — deep-link into the Web Portal	dashboardUrl

On the roadmap

These capabilities are designed and coming in a future release. No version promises.

• Test Lists (suites) — create and run named collections of tests spanning multiple projects
• Schedule and monitoring reads — read scheduled run history and monitoring status from the CLI
• Project-level test reports — aggregate pass/fail summary across all tests in a project

Feedback

The CLI gets better when you share what’s rough. Join the community to tell us what broke, what was confusing, or what you wished the CLI could do — your feedback directly shapes the roadmap.

Where to Go Next

Overview

What the CLI is and how it fits alongside the Web Portal and MCP

Quickstart

Install, configure, and run your first test in minutes

Command Reference

Every command, flag, and output shape in one place

Agent Integration

How coding agents use the CLI as a verification loop