Collect responses

You will have a page that collects answers and a place to read them.

One policy

The agent publishes one policy with the artifact. Here is a small example; each field's description explains the choice in place.

{
  "policyVersion": 1,
  "writes": "append-only",
  "perUser": "single",
  "read": "own",
  "retentionDays": null,
  "recordCap": { "perVisitor": 1, "perArtifact": 500 },
  "fields": [
    {
      "name": "attending",
      "type": "boolean",
      "required": true,
      "pii": "none",
      "description": "A yes or no answer, required before submission."
    },
    {
      "name": "name",
      "type": "text",
      "required": false,
      "pii": "contact",
      "description": "A person's name is contact data."
    }
  ]
}

own lets a visitor read their own answers through the artifact. The owner can read the collected answers with their agent. Fields can be added later; previously collected fields keep their meaning.

Read, export, erase

The Data tab shows responses and the schema to people allowed to work with the data. The owner sees raw answers. The agent can download an export through the responses export endpoint; the export follows its access, including any masking. super_data lets the owner's agent read answers back.

Erasure removes stored responses, rather than only hiding a table row. A visitor can erase their own data; the owner can erase a named response. Exported copies are separate files.

Size, field, retention and row ceilings live in Limits.

Check it worked

A submitted answer appears on Data. The export includes the response. After erasure, reading it back no longer returns that record.

Next: Collections, insights and dashboards.