Implementation brief

## Who Uses It
A customer success manager prioritizing accounts that need attention.

## What You Will Build
Build an agent that combines product usage, support history, and renewal dates into account health summaries. Recommend evidence-backed follow-up plans and track human-approved actions without inventing retention outcomes.

## Inputs
- Usage history
- Support tickets
- Renewal dates and intervention policies

## Workflow
1. Join account usage, support, and contract data.
2. Check data freshness and instrumentation health.
3. Explain account risks and propose relevant follow-up actions.
4. Build a prioritized customer-success review queue.

## Interface
An account health dashboard with trend charts, supporting events, renewal dates, and editable follow-up plans.

## Working Demo
Prioritize an account review for A1 and an instrumentation investigation for A2, with separate explanations and no automatic discount.

## Deliverables
- A working local application with the specified review interface.
- Source code, dependency lockfile, and setup commands.
- A recorded end-to-end demo using the supplied starter data.
- Example outputs, action logs, and automated tests.

## Starter Data
These fictional records are a small development fixture, not a production dataset. Implement the mock services described in the brief.

```json
{
  "accounts": [
    {
      "id": "A1",
      "weekly_active_users": [
        100,
        90,
        30
      ],
      "collector_healthy": true,
      "open_incidents": 2,
      "renewal_days": 14
    },
    {
      "id": "A2",
      "weekly_active_users": [
        100,
        0,
        0
      ],
      "collector_healthy": false,
      "open_incidents": 0,
      "renewal_days": 60
    }
  ],
  "policy": {
    "automatic_incentives": false
  }
}
```

## Constraints
Use synthetic records and draft actions. Do not contact customers or offer incentives.
Choose an available model that supports your implementation. Record its identifier and version.

## Evaluation
Review the complete application, not only the starter-case answers.
- The account summaries cite the supplied events.
- A2 is not labeled disengaged from missing telemetry alone.
- No retention improvement is claimed without outcome data.
Add one normal case, one incomplete-input case, and one failure-recovery case.