Coding Agent Integration

Install skills so your coding agent drives the TestSprite loop on its own — create, run, read failures, fix, and rerun.

What the skills do

testsprite agent install writes three skill files into your repository:

• testsprite-verify — teaches your coding agent to run the verification loop: create a test from a plan, run it and wait for the verdict, on failure pull the failure bundle, fix the code, and rerun.
• testsprite-onboard — guides first-time setup in a repo (testsprite setup, first project and test).
• testsprite-ci — sets up the pull-request gate: reads how the repository deploys previews, runs testsprite ci init github, finishes the workflow, sets the secret and verifies it on one real pull request. See CI/CD.

agent install only reads and writes local files: no network, no credentials.

Install the skills once per project with testsprite setup --agent AGENT or testsprite agent install --target AGENT.

Install the skills

testsprite agent install --target claude

Run it from your project root.

Flag	Description
--target <t>	claude, cursor, cline, antigravity, kiro, windsurf, copilot, codex. Comma-separated or repeated. Detected if omitted.
--skill <name>	testsprite-verify, testsprite-onboard, testsprite-ci. Default: all three.
--dir <path>	Project root to write into (default: current directory).
--force	Overwrite an existing skill file; a .bak backup is kept. For codex, only the managed section is replaced.

Targets can also be positional: testsprite agent install cursor codex. If a destination file already exists and differs, the install is blocked until you pass --force.

Without --target, the CLI uses the calling agent (recognized only for CLAUDECODE and CURSOR_AGENT) plus every agent whose configuration the project already has. Interactive terminals confirm the detected set; elsewhere it installs for the detected set, or claude if none, and prints an [info] line to stderr. Configuration the CLI wrote itself never counts as evidence, and an explicit --target always wins.

Supported agents

Target	Status	Mode	Landing path
claude	GA	own-file	.claude/skills/{skill}/SKILL.md
antigravity	experimental	own-file	.agents/skills/{skill}/SKILL.md
cursor	experimental	own-file	.cursor/rules/{skill}.mdc
cline	experimental	own-file	.clinerules/{skill}.md
kiro	experimental	own-file	.kiro/skills/{skill}/SKILL.md
windsurf	experimental	own-file	.windsurf/rules/{skill}.md
copilot	experimental	own-file	.github/instructions/{skill}.instructions.md
codex	experimental	managed-section	AGENTS.md

{skill} is testsprite-verify, testsprite-onboard or testsprite-ci. For codex, all installed skills go into one managed section of AGENTS.md; your other content is untouched. testsprite agent list prints this table from the CLI.

Checking skill health

testsprite agent status [--dir <path>]

State	Meaning
ok	Matches what this CLI version would install
stale	Installed by an older CLI version
modified	Edited by hand after install
unmarked	Predates the provenance marker
absent	Not installed for this target/skill
corrupt	codex only: malformed managed-section sentinels

It exits 1 when any state other than ok or absent is found, so it can gate CI.

One-shot onboarding

testsprite setup --agent AGENT

setup configures your API key, verifies your identity, and installs the skills. It takes exactly one target; without --agent it installs for every detected agent, or Claude Code if none. For several agents, use agent install. See Installation for the non-interactive --from-env --yes form.

What the agent runs

testsprite test create --project proj_8f0f6 --type frontend \
  --plan-from ./checkout-flow.plan.json --run --wait --output json

mkdir -p .testsprite
testsprite test failure get test_3a9f21c7 --out ./.testsprite/failure

testsprite test rerun test_3a9f21c7 --wait   # billed the same as a fresh run

Exit 0 means the test passed and is banked. Frontend reruns have auto-heal on by default; --no-auto-heal replays the saved script exactly.

Where to Go Next

Quickstart

The full create, run, fix loop

CI/CD

Gate your pipeline on results

Running Tests

Flags and modes for runs

Command Reference

Every command and flag