Line CLI's verification step keeps failing

Before finishing, Line runs a check: your verify_command, otherwise your tests, otherwise the type checker. Failures go back to the agent for up to 2 rounds.

If verification fails for reasons unrelated to the task:

• Wrong command detected: set the right one in .line/settings.json: {"verify_command": "npm run test:unit"}.
• Tests already failing before the change: tell Line ("tests in legacy/ already fail; ignore them") or fix them first.
• Flaky or slow checks: "verify": "nudge" asks the agent to verify instead of Line running it; "off" turns it off. For one run: --verify off.