| Zone | Where | What it holds |
|---|
| Agent rail | Left edge, vertical strip | One avatar per agent. Assistant is the default. Click to switch agents. |
| Chat | Center | Header (agent + conversation + current model), message area, input box. |
| Dock row | Below the input box | New conversation, online status dot, Skills shortcut, Tools shortcut, Commands shortcut, model picker. |
| Navigation rail | Far left, narrow icon strip | Panels: Tools, Skills, Memory, Channels, Cron, Logs, Usage, Settings. |
| Property | Value |
|---|
| Location | Dropdown chip in the dock row, just below the input box |
| Display | Current model name (for example, Claude Opus 4.6) |
| Contents | Curated short list, grouped by provider |
| Default | First provider key you set up during onboarding |
| Off-list models | Type provider-prefixed name directly into the picker |
Worked example:
Type the pattern into the picker and Clawless uses that model even when it is not on the visible short list. Use this only when you have a specific reason; the short-list defaults handle almost everything.
| You see | You do | What it means |
|---|
| Empty input | Click into it, type, press Enter | Sends the message. |
| Status line below your message | Read it | Activity indicator: thinking, reading a file, searching the web, calling a tool. |
| Reply streaming word by word | Read along | Streamed output. No need to wait for the whole reply. |
| Square icon where Send used to be | Click it | Stop. Halts the stream. Usually instant. |
| Red banner with Force Stop | Use as last resort | Stop did not take effect after about ten seconds. |
| Code block in reply | Click Copy in the corner | Lifts the block to clipboard. |
| Rule | Detail |
|---|
| No per-conversation lock | Switch anytime. The new model sees the full conversation history. |
| Cost meter follows current model | Only the message you are about to send is billed at the new model’s rate. Past messages stay billed at their original model. |
| Common pattern: start big, then small | Use a capable model for the hard first question, switch to cheaper for follow-ups. Drops cost per message 5x to 10x without restarting. |
| Reverse also works | Start small for an easy task, switch up when it gets harder. |
| What you click | What changes |
|---|
| A different agent on the left rail | Conversation swaps (each agent has its own thread). Model picker updates to that agent’s preferred model. Neither thread is lost. |
| Picker on the current agent | Model changes for this conversation only. Permanent default lives on the agent’s settings (covered in a later lesson). |
The model name in the conversation header shows the current model you have selected. If you switch models mid-conversation, the header reflects the model you have selected.
| Surprise | Why it happens |
|---|
| First message of a session feels slow | Engine warmup plus possible provider cold start. Subsequent messages are faster. |
| Status line appears below your message | That is the activity indicator. Tells you what the agent is doing. Not a problem. |
| Switching agents loads a different conversation | Each agent has its own thread, by design. Stay in one agent to continue the same topic. |
| Picker shows different defaults across agents | Each agent has its own preferred model. Picker is following the agent. |
The Usage dashboard in the navigation rail, after your first few conversations, breaks down cost by provider, model, and conversation. Reading the real numbers is more durable than guessing what each model costs you.
- Treating the model choice as a permanent setup decision (it is a live dial)
- Re-pasting context after switching models (the new model already has it)
- Starting every conversation on the biggest model (cheap models handle a lot of follow-up work fine)
- Typing slash commands by accident on the first day (safe to ignore; covered later)
A later lesson covers connecting providers (API keys vs ChatGPT OAuth), so you understand what is in the picker and why. A later lesson covers Tools, Skills, Memory, Channels, Cron, Logs, Usage, and Settings, one by one. Everything else in the track sits on top of the model picker you just met.
