We're launching OpenClaw LLM Proxy on Tuesday at 12:01 AM PDT - a drop-in OpenAI API replacement with built-in PII scanning and injection detection.
Curious to hear from the community before we go live: when you send prompts to OpenAI/Anthropic/Google, is there anything scanning for sensitive data before it reaches the provider?
OpenClaw sits between your app and any LLM provider. Every prompt is scanned for PII (SSN, credit cards, API keys) and injection attacks before reaching the model. Every response is scanned for data leaks.
Routes across 6 backends (OpenAI, Anthropic, Google, Ollama, vLLM) by model prefix. Block, redact, or log - your choice. Multi-tenancy with per-team budgets. Spend tracking. Response caching. Prometheus metrics. Helm chart.
113 tests. One Docker command.
