The AI is not responding or shows an error

If the chat never answers, answers with an error, or stops halfway, the cause is almost always one of these.

1. The key is missing, wrong, or expired

Symptoms: "unauthorized", "invalid key", 401, or 403.

Fix: Run Bricklayer: Choose AI Agent (Ctrl+I), choose your provider again, and re-enter the key or sign in again. If you recently created a new key, make sure you pasted the new one, all of it.

2. The model is unavailable

Symptoms: "model not found", "no endpoints", or a model that worked yesterday suddenly fails.

Fix: Free models go offline without notice. Pick another one from the OpenRouter model guide.

3. You hit a usage limit

Symptoms: 429, "rate limited", "usage limit", or "quota".

Fix: See "Rate limited" or 429 errors.

4. Your agent is not signed in

Symptoms: Claude Code, Codex, or another installed agent never starts, or asks you to log in.

Fix: Sign in to the agent itself. For Codex on a free ChatGPT account, choose Sign in with ChatGPT, not the API key option.

5. The provider is having an outage

Symptoms: Everything is set up correctly, other models fail too, and nothing changed on your side.

Fix: Check the provider's status page, or switch providers for now.

When you ask for help

Include the exact error text and which AI and model you use. Never include the key itself. If you posted a key by accident, revoke it right away.