Gateway Proxy
Route LLM traffic through the gateway for zero-code governance.
The Gateway Proxy is the simplest integration. Point your existing LLM client at the Palveron gateway URL and authenticate with your Palveron project key — no code changes needed.
OpenAI (Node.js)
const openai = new OpenAI({
baseURL: 'https://gateway.palveron.com/v1',
apiKey: 'pv_live_abc123...',
});The Authorization header carries your Palveron key. The upstream LLM provider and its key come from your project configuration (Settings → Neural Gateway): your BYOM key (or a custom endpoint) — never from the request itself. Blocked prompts return a governance error instead of the LLM response; allowed prompts are forwarded (masked if a MODIFY policy fired).
Optionally attribute proxy traffic to a registered agent with an X-Palveron-Agent header.
Fail behavior
The proxy itself is fail-closed: if the gateway cannot record the governance decision, the request does not reach the LLM. Client-side fail-open ("pass through when the gateway is unreachable") is a feature of the SDK framework adapters (fail_open option), not of the raw proxy.