Troubleshooting

My local model does not appear

• Confirm the server is running. For Ollama, `ollama list` should respond.
• Check the host in Settings → Local models. Ollama defaults to 127.0.0.1:11434, LM Studio to 127.0.0.1:1234.
• Make sure the model is actually pulled, not just named. Pulling multi-gigabyte weights takes a while.

Generation is very slow

• The model is probably spilling out of VRAM into system RAM. Drop to a smaller size and compare.
• Check the fit rating for your GPU. 'Tight' means it runs but with little headroom.
• Close other GPU-heavy applications, including browsers with hardware acceleration.

A cloud model says to ask an admin

That means no key is configured for that provider. There is deliberately no paste field in the product — an administrator configures cloud providers, or your plan covers the proxied ones. Local models are unaffected.

The agent broke something

Use code restore to roll the working tree back to a checkpoint in the session, or git checkout the file. Committing before a large agent run is the cheapest insurance there is.