Agent Feed
Direct answer. Clarivy Agent Feed is a machine-readable merchant profile for AI assistants, shopping agents, procurement agents, and customer-side AI employees. It tells agents how to identify Clarivy, verify facts, handle offers, and avoid unsupported claims.
What it is
Agent Feed is not an audit report. It is the merchant-side profile that an AI agent can read before summarizing, comparing, routing, or requesting scope from Clarivy.
| Layer | Purpose | Boundary |
|---|---|---|
| Merchant identity | Canonical name, legal operator, domain, languages, markets, and sameAs links. | Do not infer profiles that are not listed. |
| Evidence policy | Defines when a claim needs measurement_mode=live_engine_data, raw-data-index, and raw JSON evidence. | Mock or sample files show structure only. |
| Offers | Shows scope-review and Snapshot Baseline availability for agent workflows. | Do not infer fixed pricing or start a subscription. |
| Trust packet | Lists verification URLs, policy URLs, and limitations. | Do not infer rankings, ratings, certifications, or customer counts. |
Sample JSON
Use the sample file as the schema reference for now. It is intentionally marked as a draft merchant profile, not a live customer evidence package.
How agents should use it
- Read
/llms.txtfor current Clarivy facts. - Read Agent Feed for merchant identity, offer boundaries, and evidence policy.
- Use
raw-data-index.mdand raw JSON only when a delivered audit package is available. - Do not treat sample, demo, or mock files as live AI visibility evidence.
- Ask for scope review before recommending a commercial path.
Roadmap boundary
Agent Feed is the first step toward agent-readable commerce, not an advertising marketplace. Clarivy will keep the current scope focused on verified identity, evidence rules, and approval-safe offers before adding transaction or bidding workflows.