Protowall is the gateway between your application and foundation models. Inspect every request, enforce policy, and route across providers from a single SDK.
// Import the Protowall SDK, that's it.
import Protowall from '@protowall/sdk';
const client = new Protowall({
apiKey: process.env.PROTOWALL_API_KEY,
});
// All requests now flow through Protowall
const response = await client.chat.completions.create({
messages: [{ role: 'user', content: prompt }],
});Detect prompt injection, jailbreaks, and sensitive data exposure before a request reaches your model. Define policies that block, flag, or allow.
Route requests across providers by cost, latency, or capability. Automatic failover keeps your application available when upstream providers degrade.
Full request and response tracing with cost attribution by user, key, or model. Debug production issues without leaving the dashboard.
A single interface to OpenAI, Anthropic, Google, Mistral, and more. Consistent schemas, unified logs, and normalized error handling across providers.
Granular limits per key, user, or workspace. Enforce budgets and receive alerts before anomalous usage becomes an incident.
Issue scoped virtual keys to teams and services. Revoke instantly, audit per-key usage, and keep provider credentials off your application.
Drop in our SDK for Python, Node, or Go, or call the REST API directly. No application rewrites required.
Configure detection, routing, and rate-limit policies from the dashboard or as code. Changes take effect instantly.
Every request is inspected, routed, and logged. Your team gets security, observability, and cost control from day one.
Protowall is built on a zero-trust architecture. Prompts and completions are never persisted by default. We operate on metadata so your sensitive data stays in your environment.
We're onboarding design partners ahead of general availability. Join the private beta or talk to the team about your use case.