Request Parameters

Overview

request_params is a route-local plugin that validates and trims OpenAI Chat Completions request bodies before they are forwarded to backends.

Key Advantages

• Caps expensive parameters (max_tokens, n) per route.
• Blocks sensitive parameters such as logprobs / top_logprobs for tiers that should not expose token distributions.
• Optionally strips unknown top-level JSON fields to reduce surprise passthrough behavior.

What Problem Does It Solve?

Model routing can restrict which backend serves a request, but clients can still amplify cost or extract logits via request parameters. This plugin enforces per-decision limits on the request body after a decision matches.

When to Use

• a tier or route must not request logprobs or multiple completions
• you need a hard ceiling on max_tokens or n independent of client input
• unknown JSON fields should not be forwarded to backends

Configuration

Add the plugin under routing.decisions[].plugins:

plugins:
  - type: request_params
    configuration:
      blocked_params:
        - logprobs
        - top_logprobs
      max_tokens_limit: 500
      max_n: 1
      strip_unknown: true

In DSL, the same plugin can appear as:

PLUGIN request_params {
  blocked_params: ["logprobs", "top_logprobs"]
  max_tokens_limit: 500
  max_n: 1
  strip_unknown: true
}

This plugin enforces a bounded set of OpenAI Chat Completions fields. It is not a general JSON-schema firewall and does not authorize a caller. Test strip_unknown against clients that add provider-specific fields before enabling it. See a complete example: config/fragments/plugin/request-params/budget-tier.yaml.