# Interaction Telemetry Guardrails

## Objective

Collect useful interaction signal without recording what people type or read.

Redaction, sampling, and consent rules applied to interaction events before they leave the browser.

## Before You Begin

This feature is being added to an application that already exists and already
works. Do not scaffold a new project, and do not assume a blank slate.

Inspect the codebase first and establish:

- The existing application structure and where code of this kind already lives.
- The framework and version in use.
- The existing design system — colours, spacing, typography, and component conventions.
- Existing UI components you can reuse instead of writing new ones.
- The existing database structure, if this feature needs to persist anything.
- The existing authentication and authorization system, if this feature is user-scoped.
- Dependencies already installed, so you don't add a library that duplicates one.
- The existing test setup and conventions.

Only start writing code once you understand the above. If the application
already implements part of this feature, extend it rather than replacing it.

## Implementation Instructions

1. Set the default to sending no content. An event records that a thing was interacted with, its type and its identifier; it does not record the text inside it unless that field is explicitly listed as safe.
2. Redact at the point of collection rather than after ingestion. Once a customer's message body reaches the analytics vendor, deleting it is a support request to a third party and a disclosure you have already made.
3. Sample the high-frequency signals — scrolling, hovering, typing, resizing — at a rate recorded on the event itself, so the numbers can be scaled back up correctly instead of being read as raw counts.
4. Honour the user's consent choice and the workspace's privacy configuration before anything is queued, and drop the queue when consent is withdrawn rather than sending what was already collected.
5. Distinguish an action a person took from a change the interface made. Restored scroll positions, programmatic focus, prefetches, retries, and automated tests all produce interaction-shaped events that are not interactions.

## UI and UX Requirements

Match the application's existing design system exactly. Reuse its components,
spacing, and typography. This feature should look like it was always there.

## Responsive Requirements

Works on mobile, tablet, and desktop. Touch targets are large enough to hit on a
phone, and nothing overflows horizontally at 320px.

## Accessibility Requirements

- Fully keyboard navigable.
- Correct semantic elements and ARIA roles.
- Visible focus states.
- Meets WCAG AA contrast.
- Dynamic changes are announced to screen readers.
- Respects prefers-reduced-motion.

## Edge Cases

- Private content leaks through the least obvious properties: a page title containing a customer name, a URL path with an email in it, a form label carrying a filename. Redaction must cover titles, paths, labels, and error text, not only field values.
- High-frequency events dominate the volume and the cost while carrying little information per event. Sample them, and record the sampling rate on the event so nobody later reports a tenth of the real number as fact.
- Consent and workspace privacy settings are checked before an event is queued, not before it is sent. A queue built up in advance and flushed after a refusal is exactly the outcome consent was supposed to prevent.
- Events generated by the app rather than the user — restored state, automated retries, synthetic monitoring, test runs — must be excluded or clearly marked, or engagement metrics will measure the software's own behaviour.
- The volume of events must be bounded so a stuck component cannot flood the queue and the network from a single session.
- Some jurisdictions and enterprise workspaces require regional storage or complete opt-out, so the destination and the ability to send nothing at all must both be configurable.
- A retention period has to be set deliberately for interaction data, because the default in most tools is longer than any legitimate use of it.

## Testing

Exercise the feature end to end in the running application. Cover every edge case
above, then run the existing test suite and confirm nothing regressed.

## Acceptance Criteria

- [ ] Events carry no user-entered text, and safe fields are enumerated explicitly rather than assumed.
- [ ] Redaction happens in the browser before any event is transmitted.
- [ ] High-frequency events are sampled and each event states the rate it was sampled at.
- [ ] Consent and workspace privacy are evaluated before an event is queued, and withdrawal drops the pending queue.
- [ ] App-generated interactions are excluded or distinctly marked.
- [ ] Per-session event volume is capped.
- [ ] Destination region, opt-out, and retention period are configurable and documented.
- [ ] The feature matches the existing design system.
- [ ] No existing functionality is broken.

## Adaptation Rules

- Match the existing design system. Do not introduce a new colour palette,
  spacing scale, or component library.
- Reuse existing components and utilities wherever they fit.
- Follow the naming, file layout, and code style already present.
- Do not upgrade, replace, or remove existing dependencies to make this
  feature fit. Adapt the feature to the app, not the app to the feature.
- Do not break existing functionality. If a change is genuinely required in
  existing code, make the smallest one that works and say so.
- If something in these instructions conflicts with how the application is
  built, follow the application and explain the deviation.

## Final Verification

Before you report the work as done:

1. Re-read the acceptance criteria above and check each one against what you
   actually built.
2. Run the application and exercise the feature end to end.
3. Run the existing test suite and confirm you have broken nothing.
4. Check the feature on mobile, tablet, and desktop widths.
5. Check keyboard navigation and focus handling.
6. Summarize what changed: files added, files modified, and anything you
   deliberately did differently because of how this application is built.

If any acceptance criterion is unmet, fix it before reporting completion.