Skip to content
Merged
2 changes: 1 addition & 1 deletion docs-site/index.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ summary: "ClawBox is pre-configured AI hardware that runs OpenClaw or Hermes Age
</p>

<Tip>
📺 **Hands-on walkthrough (English voice-over)** — *"ClawBox: A Ferramenta que Muda Como Você Trabalha,"* including a manual ClawBox install via `installer.sh` → [watch on YouTube](https://youtu.be/KMDTF7N8uzU).
📺 **Hands-on walkthrough** — *"I Tested ClawBox — Is This the Future of Personal AI?"* by Bootable USBs → [watch on YouTube](https://www.youtube.com/watch?v=1cekuQjVikU).

📺 **Watch an independent review** — maker Kevin McAleer breaks down *"What is ClawBox, and who is it for?"* → [watch on YouTube](https://www.youtube.com/live/rOkt0Ev-dQ0).

Expand Down
10 changes: 10 additions & 0 deletions docs-site/setup/choose-ai-provider.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -46,6 +46,16 @@ Connect** (Jetson Orin Nano) is best paired with cloud providers or smaller loca
See [Hardware](/hardware/clawbox-connect).
</Note>

## Reasoning effort

You can control how hard the model thinks before answering, per conversation, from the brain icon in the chat header (or with `/think <level>`). More thinking gives better results on hard problems and costs more time and tokens.

With **ClawBox AI**, the DeepSeek-backed default, three levels are available: **Off** (the fast default), **High**, and **X-High** (maximum reasoning). Off keeps everyday replies quick; you opt in to reasoning when a task needs it. Other providers you connect may offer their own set of levels.

<Note>
For the technical detail of how these levels are configured — and why they are set in code rather than by hand — see [AI Providers → Reasoning effort](/technical/ai-providers#reasoning-effort).
</Note>

Comment thread
coderabbitai[bot] marked this conversation as resolved.
Resolved
## Switching later

You can change provider any time from the device settings — no re-setup required.
22 changes: 22 additions & 0 deletions docs-site/technical/ai-providers.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -44,6 +44,28 @@ Some native OpenClaw provider plugins expect auth stores ClawBox doesn't populat
- Requests proxy through the portal's AI endpoint; both models advertise reasoning support including **X-High** effort (mapped upstream to DeepSeek's maximum reasoning mode).
- Changing the linked portal account automatically unpairs ClawKeep (backups belong to the account).

### Reasoning effort

ClawBox AI Flash and ClawBox AI Pro expose three reasoning-effort levels, selectable per conversation from the brain icon in the chat header (or with `/think <level>`):

| Level | Behavior |
|---|---|
| `off` | No extended reasoning — the default. The cloud proxy disables thinking unless you opt in, keeping replies fast and cheap. |
| `high` | Extended reasoning for harder tasks. |
| `xhigh` | Maximum reasoning. Mapped upstream to DeepSeek's `reasoning_effort: "max"`. |

The chat header may show more levels globally, but for the ClawBox AI (DeepSeek) provider only these three take effect; anything else is clamped or rejected by the gateway.
Comment thread
coderabbitai[bot] marked this conversation as resolved.
Resolved

<Warning>
**This list is set programmatically, not by hand-editing config.** Three mechanisms each independently overwrite a manual edit, so all three must be satisfied in code:

1. **`gateway-pre-start.sh` re-asserts it on every gateway start.** The boot script rewrites `compat.supportedReasoningEfforts` on the DeepSeek models back to the target list each time the gateway starts. A hand edit to `openclaw.json` survives only until the next restart.
2. **`npm install -g openclaw@latest` overwrites the plugin catalog.** The levels are also declared in the OpenClaw package at `dist/extensions/deepseek/openclaw.plugin.json`, which every system update replaces wholesale. ClawBox re-applies the patch on each boot precisely because of this.
3. **A configured provider overrides the plugin catalog entirely.** When `openclaw.json` defines the `deepseek` provider, OpenClaw ignores the plugin's `modelCatalog` for those models — so the `compat` block must live on the configured entry (written by the setup route), not on the plugin file.

Without the declaration, OpenClaw's `catalogSupportsXHigh()` returns false, `sessions.patch` rejects `xhigh` ("use off|minimal|low|medium|high"), and the chat header's **X-High** option silently fails — even though the transport layer already translates OpenClaw `xhigh` → DeepSeek `reasoning_effort: "max"` correctly. The capability is present; only the catalog declaration was ever missing. The two source-of-truth locations are `src/app/setup-api/ai-models/configure/route.ts` (written at pairing) and `scripts/gateway-pre-start.sh` (enforced at every boot).
</Warning>

## Local models (Ollama / llama.cpp)

- Local lanes authenticate with a **per-install bearer token** through a local proxy — nothing is exposed unauthenticated, even on localhost.
Expand Down
Loading