@insightlabai.dev/cli

@insightlabai.dev/cli

Find UX bugs before your users do. Points synthetic users at a URL, has them attempt a task, and returns severity-ranked findings.

npm install -g @insightlabai.dev/cli

Setup

Create an API key at insightlabai.dev/demo, then:

export INSIGHTLAB_API_KEY=il_live_...

Usage

insightlab run https://staging.acme.com \
  --task "sign up and reach the dashboard" \
  --audience "first-time users, low patience" \
  --agents 5

5 agents · 1m 48s · 3 issues found

CRITICAL  Signup CTA sits below the fold at 1280x800  (3 of 5)
          → Move the primary CTA above the fold at common viewport heights.

MAJOR     Password rules only appear after a failed submit  (4 of 5)
          → Show the password requirements before the first submit.

In CI

--fail-on makes the command exit non-zero when a finding lands at or above that severity, so a usability regression can hold a merge the same way a failing test does.

insightlab run "$PREVIEW_URL" --task "complete checkout" --fail-on critical

Runs take minutes rather than seconds, so point it at the flows that matter most rather than every route.

Options

Option	Description
--task <text>	What a user should accomplish. Required.
--audience <text>	Who is attempting it. Different users fail in different places.
--agents <n>	How many synthetic users to run. Default 5.
--fail-on <level>	Exit 1 if a finding is at or above critical, major, or minor.
--json	Emit raw JSON on stdout. Progress goes to stderr, so it stays pipeable.

Environment

Variable	Description
INSIGHTLAB_API_KEY	Required.
INSIGHTLAB_API_URL	Override the API base URL. Defaults to https://insightlabai.dev.

Exit codes

Code	Meaning
0	Run finished; no finding reached the --fail-on threshold.
1	A finding reached the threshold.
2	The command could not run: bad usage, bad key, or the API was unreachable.