When the Store API Goes Dark: Designing a Trustworthy Continuity Layer for Agentic Commerce

Description and use

When a store’s REST API goes down, an AI agent must keep discovering merchants by serving the last verified catalog snapshot, ensuring merchant control and data freshness.

Problem

An AI shopping agent must reliably discover merchants even when a store’s REST API is temporarily unavailable, without penalizing merchants for unrelated outages or serving stale data as authoritative.

Outcome

The continuity layer achieved a 7.6‑second source switch, zero snapshot requests to the merchant during outage, and zero AI calls during recovery in a controlled test, while maintaining merchant control and data freshness.

Workflow

• 1. Merchant publishes signed snapshot via Bridge plugin. 2. Network receives heartbeat and stores hot candidate. 3. On live API failure, network switches to hot snapshot locally. 4. Snapshot is served until freshness expires or revocation occurs. 5. When API recovers, network performs full authoritative re‑read before returning to live.
• 6. Fail‑closed logic distinguishes outages, revocations, and missing plugins.
• 7. Operational checks prevent flapping and ensure concurrency safety.

Evidence

The KaliCart Bridge plugin (WordPress/WooCommerce) periodically publishes a static snapshot of the store's public catalog: a manifest plus content shards.

Source

When the live API fails, switching source is a local operation. In our canary on a test store with 110 products: the source switch completed in 7.6 seconds from decision to serving; the merchant's server received zero snapshot requests during the switch; zero AI calls were made.

Source

We made a hard rule: switching between live and snapshot is deterministic and AI‑free.

Source