OpenClaw
Run PromptShield as the gateway between OpenClaw and your LLM provider.
OpenClaw supports custom providers via openclaw.json. Point it at PromptShield and every request gets scanned, rate limited, and logged before it reaches the LLM.
OpenClaw → PromptShield (:8080) → Anthropic / OpenAI / GeminiConfigure the gateway
Set your provider and key in .env:
PROMPTSHIELD_PROVIDER=anthropic
ANTHROPIC_API_KEY=sk-ant-api03-xxxxFor multi-provider routing, see Providers.
Then start the gateway:
make runConfigure OpenClaw
Edit ~/.openclaw/openclaw.json:
{
"models": {
"providers": {
"promptshield": {
"baseUrl": "http://localhost:8080/v1",
"apiKey": "local",
"api": "openai-completions"
}
}
},
"agents": {
"defaults": {
"model": {
"primary": "promptshield/claude-sonnet-4-5"
}
}
}
}apiKey is a placeholder; the gateway holds the real upstream key. Copy from examples/openclaw.json.