@briefcase-ai/controls

Controls layer for AI invocations: ports (quota, entitlements, trace export), a gateway that enforces hard caps, quota, and throttle cooldowns around every call, a suggestion pipeline, throttle classification and retry helpers, and decision-record exporters. Bedrock-specific helpers (model-id resolution, capability probes, pricing, credential provider) live on the ./bedrock subpath.

Install

npm install @briefcase-ai/controls

Usage

import { MemoryQuotaStore, createGateway } from "@briefcase-ai/controls";

const gateway = createGateway({
  quotaStore: new MemoryQuotaStore(),
  buckets: {
    suggestions: { capacity: 10, refillSecondsPerToken: 60 },
  },
});

const outcome = await gateway.invoke({
  tenantId: "tenant-1",
  bucket: "suggestions",
  fn: () => callYourModel(),
});

if (outcome.ok) {
  console.log(outcome.value, outcome.tokensRemaining);
} else {
  console.log(outcome.reason, outcome.cooldownUntil);
}

MemoryQuotaStore is in-process; production stores implement the QuotaStore port.

Bedrock subpath

@briefcase-ai/controls/bedrock exports model-id resolution, Converse capability probes, list-price lookup, and a memoized credential provider; the @ai-sdk/amazon-bedrock and @aws-sdk/credential-providers peer dependencies are optional and only needed by the credential provider and adapters.

Source: https://github.com/briefcasebrain/briefcase-ai-sdk