From 0c74e70f65c9622f2045b93e72d51cf38605904b Mon Sep 17 00:00:00 2001
From: Shin <shin-bot-litellm@berri.ai>
Date: Sun, 1 Feb 2026 21:04:50 +0000
Subject: [PATCH] docs: add OpenClaw integration tutorial

Shows how to route OpenClaw through LiteLLM proxy for:
- Unified logging and cost tracking
- Virtual keys with spend limits
- Model switching without config changes
- Remote access via Gateway mode
---
 docs/my-website/docs/tutorials/openclaw.md | 141 +++++++++++++++++++++
 docs/my-website/sidebars.js                |   1 +
 2 files changed, 142 insertions(+)
 create mode 100644 docs/my-website/docs/tutorials/openclaw.md

diff --git a/docs/my-website/docs/tutorials/openclaw.md b/docs/my-website/docs/tutorials/openclaw.md
new file mode 100644
index 00000000000..a89b192f1ca
--- /dev/null
+++ b/docs/my-website/docs/tutorials/openclaw.md
@@ -0,0 +1,141 @@
+# Using OpenClaw with LiteLLM
+
+Run [OpenClaw](https://openclaw.ai) — the open-source AI assistant — through your LiteLLM proxy. Get unified logging, cost tracking, and model management while keeping OpenClaw's full capabilities.
+
+## Why use LiteLLM with OpenClaw?
+
+- **Unified API** — Route OpenClaw through any model provider (Anthropic, OpenAI, Azure, Bedrock, etc.)
+- **Cost tracking** — Track spend across all your AI tools in one place
+- **Virtual keys** — Give OpenClaw its own API key with spend limits
+- **Model fallbacks** — Automatic failover if your primary model is down
+
+## Quick Start
+
+LiteLLM's native `/v1/messages` endpoint speaks Anthropic's protocol — perfect for OpenClaw.
+
+### 1. Start LiteLLM Proxy
+
+```bash
+# Set your API key (works with any provider LiteLLM supports)
+export ANTHROPIC_API_KEY="sk-ant-..."
+
+# Start the proxy
+litellm --model anthropic/claude-sonnet-4-5-20250929
+
+# Running on http://0.0.0.0:4000
+```
+
+### 2. Configure OpenClaw
+
+Point OpenClaw to your LiteLLM proxy:
+
+```bash
+# Set environment variables
+export ANTHROPIC_API_BASE="http://localhost:4000"
+export ANTHROPIC_API_KEY="sk-anything"  # or your LiteLLM virtual key
+
+# Start OpenClaw
+openclaw
+```
+
+That's it! OpenClaw now routes through LiteLLM via the native `/v1/messages` endpoint.
+
+### Alternative: Config file
+
+Add to your OpenClaw config (`~/.openclaw/config.json5`):
+
+```json5
+{
+  env: {
+    ANTHROPIC_API_BASE: "http://localhost:4000",
+    ANTHROPIC_API_KEY: "your-litellm-key"
+  },
+  agents: {
+    defaults: {
+      model: { primary: "anthropic/claude-sonnet-4-5-20250929" }
+    }
+  }
+}
+```
+
+## Using Virtual Keys
+
+Want to track OpenClaw's spend separately? Create a virtual key:
+
+```bash
+# Generate a key for OpenClaw
+curl -X POST 'http://localhost:4000/key/generate' \
+  -H 'Authorization: Bearer your-master-key' \
+  -H 'Content-Type: application/json' \
+  -d '{
+    "key_alias": "openclaw",
+    "max_budget": 50.00,
+    "models": ["claude-sonnet-4-5-20250929", "claude-opus-4-5-20251101"]
+  }'
+```
+
+Use the returned key as `ANTHROPIC_API_KEY` in OpenClaw.
+
+## Using Different Models
+
+LiteLLM lets you swap models without changing OpenClaw config. Update your `config.yaml`:
+
+```yaml
+model_list:
+  # Map "claude-sonnet-4-5-20250929" to any model
+  - model_name: claude-sonnet-4-5-20250929
+    litellm_params:
+      model: azure/gpt-4o  # or bedrock/claude, openai/gpt-4, vertex_ai/gemini-pro, etc.
+      api_key: os.environ/AZURE_API_KEY
+      api_base: https://your-resource.openai.azure.com
+```
+
+OpenClaw requests `claude-sonnet-4-5-20250929` → LiteLLM routes to your chosen model.
+
+## Remote Access (Gateway Mode)
+
+Running OpenClaw Gateway on a server? Point it to your LiteLLM instance:
+
+```json5
+// ~/.openclaw/config.json5 on your server
+{
+  env: {
+    ANTHROPIC_API_BASE: "http://litellm-host:4000",
+    ANTHROPIC_API_KEY: "your-litellm-key"
+  },
+  gateway: {
+    bind: "lan",  // or "tailnet" for Tailscale
+    port: 4321
+  }
+}
+```
+
+## Troubleshooting
+
+### "Connection refused"
+
+Make sure LiteLLM is running and accessible:
+```bash
+curl http://localhost:4000/health
+```
+
+### "Invalid API key"
+
+If using virtual keys, ensure the key has access to the models OpenClaw uses:
+```bash
+curl http://localhost:4000/key/info \
+  -H "Authorization: Bearer your-key"
+```
+
+### Model not found
+
+Check your LiteLLM config includes the model OpenClaw is requesting:
+```bash
+curl http://localhost:4000/v1/models
+```
+
+## Learn More
+
+- [OpenClaw Documentation](https://docs.openclaw.ai)
+- [OpenClaw GitHub](https://github.com/openclaw/openclaw)
+- [LiteLLM /v1/messages docs](/docs/anthropic_unified/)
diff --git a/docs/my-website/sidebars.js b/docs/my-website/sidebars.js
index 7b8a09196ee..ae6501f9f6c 100644
--- a/docs/my-website/sidebars.js
+++ b/docs/my-website/sidebars.js
@@ -185,6 +185,7 @@ const sidebars = {
       items: [
         'tutorials/azure_openai',
         'tutorials/instructor',
+        'tutorials/openclaw',
         'tutorials/oobabooga',
         "tutorials/gradio_integration",
         "tutorials/huggingface_codellama",