🦞 OpenClaw Model Switcher

Prerequisite: OpenClaw must be installed and initialized (~/.openclaw/openclaw.json must exist)

# Clone the repo
git clone https://github.com/yuanrengu/configadpter.git
cd configadpter
# Start the tool (zero dependencies — no pip install needed)
python3 openclaw_config.py
# Browser opens automatically at http://localhost:7890

# Stop: press Ctrl+C in the terminal
# If running in background, kill the process with:
lsof -ti :7890 | xargs kill -9

Requirements: Python 3.8+, standard library only — no pip install needed.

When you apply a model switch, the tool backs up the current config first, then only modifies these 4 areas in ~/.openclaw/openclaw.json. Everything else is left untouched.

models.providers.<provider> ← Rebuilds the target provider's connection config
auth.profiles.<provider>:default ← Adds or updates the auth profile entry
agents.defaults.model.primary ← Sets "provider/modelId" (the core field OpenClaw reads)
agents.defaults.models.<primary> ← Adds the model alias (display name)

API Key handling: If the key field is left blank, the existing key is inherited — it will never be lost.

channels.* ← Telegram / Discord channel settings
gateway.* ← Port, auth token, node config
tools.* ← Tool profiles
commands.* ← CLI settings
session.* ← Session config
hooks.* ← Internal hook config
wizard.* ← Wizard state
Other providers' configurations

The version shown in the top-right corner (e.g. v2026.3.2) comes from meta.lastTouchedVersion in openclaw.json. It is written and maintained by OpenClaw itself — this tool only reads and displays it.

Usually no. The OpenClaw Gateway monitors openclaw.json for changes and hot-reloads most configuration updates automatically.

Note: When switching to a brand-new provider that was never configured before, OpenClaw may need a moment to initialize the new connection. If the new model doesn't take effect immediately, check the OpenClaw web console or restart the OpenClaw process once.

1. Click the 🔵 DeepSeek card
2. Enter your DeepSeek API Key (sk-...)
3. Select a model (e.g. DeepSeek Chat V3)
4. Click ✅ Apply

1. Click the 🌙 Moonshot (Kimi) card (shows "✓ Active")
2. Leave the API Key blank (the existing key is reused automatically)
3. Select a model and click Apply

1. Click the ⚙️ Custom card
2. Enter the Base URL (e.g. https://your-api.com/v1) and API Key
3. Type a model ID and display name, then click Add
4. Select the model and click Apply

.
├── openclaw_config.py # Backend: provider templates, HTTP server, config read/write
├── index.html # Frontend: Web UI (HTML / CSS / JS, dark theme)
├── docs/
│ ├── screenshot_home.png
│ └── screenshot_deepseek.png
└── README.md

OpenClaw paths:

• Config file: ~/.openclaw/openclaw.json
• Backup dir: ~/.openclaw/openclaw.json.bak.tool.*

• The server only listens on 127.0.0.1:7890 — never exposed to the network
• Every write is preceded by an automatic backup, recoverable from ~/.openclaw/
• API keys are never pre-filled in the form and never printed to the terminal

MIT