TestSprite/testsprite-cli
A Test CLI That Makes Agents Face Their Own Breakage
It gives coding agents a deterministic way to verify their own code against live apps and receive structured failure bundles they can act on directly.
Collecting fresh signals — velocity needs a few days of history.
collecting data…
star history
What it does
TestSprite CLI is the official terminal interface for the TestSprite cloud testing platform. It lets developers and coding agents create, run, and rerun end-to-end tests against live browsers and APIs, then pull back structured failure reports. The tool is built around a loop: describe a behavior, test it, bank the passing result, and when something breaks, hand the agent a single snapshot of the failure to fix.
The interesting bit
The CLI refuses to stitch failure data across different test runs, which means an agent never reasons over a “frankenstein context.” Instead, test failure get returns one self-consistent bundle—screenshots, DOM snapshots, a root-cause hypothesis, and a recommended fix target—all tied to a single snapshotId. That is a deliberate design choice for autonomous workflows, not just human convenience.
Key highlights
- Targets agentic workflows with first-class support for Claude, Cursor, Cline, Codex, and others via
agent install. - Runs tests against live browsers and real APIs in the cloud; no mocks.
- Returns deterministic, scriptable output with stable
--output json, predictable exit codes, and a--dry-runmode for offline exploration. - Claims to have helped the cheapest model top a public leaderboard with 89% correctness at half the cost of the priciest competitor.
Verdict
Worth a look if you are already using or evaluating TestSprite’s cloud platform and want your agent or CI pipeline to drive verification from the terminal. Skip it if you need a fully offline, self-hosted testing runner—this is a client for a commercial cloud service.