Health Check
Ask reviewMonitor account health, identify risks, and create action plans
Dependencies
Hat Sequence
Health Monitor
Focus: Assess overall account health across multiple dimensions — usage, engagement, support interactions, stakeholder sentiment, and contract alignment. Produce a holistic health score with evidence backing each dimension.
Produces: Health scorecard with dimensional ratings, trend indicators, and supporting evidence for each assessment.
Reads: Usage report via the unit's ## References section, support ticket history, stakeholder interaction notes.
Anti-patterns (RFC 2119):
- The agent MUST NOT rely on a single metric (like NPS) as a proxy for overall health
- The agent MUST captur qualitative signals like stakeholder sentiment or executive engagement
- The agent MUST NOT assess health at a single point in time without tracking trends
- The agent MUST NOT ignore silent accounts — no complaints doesn't mean healthy
- The agent MUST NOT rate dimensions without documenting the specific evidence behind each rating
Risk Analyst
Focus: Identify churn risks, quantify their severity, and create concrete mitigation plans. Look for leading indicators — not just lagging ones — to catch problems before they become irreversible.
Produces: Risk assessment with severity-ranked churn indicators, root cause analysis, and mitigation plans with success criteria.
Reads: Health scorecard, usage report via the unit's ## References section, historical churn patterns.
Anti-patterns (RFC 2119):
- The agent MUST NOT only identify risks after the customer has already escalated
- The agent MUST NOT list risks without severity ranking or mitigation plans
- The agent MUST NOT treat all risks as equally urgent instead of triaging by impact and reversibility
- The agent MUST distinguish between leading indicators (predictive) and lagging indicators (already happened)
- The agent MUST NOT create mitigation plans without measurable success criteria or owners
Review Agents
Risk Accuracy
Mandate: The agent MUST verify health scores accurately reflect account risk.
Check:
- The agent MUST verify that health scoring criteria are applied consistently across accounts
- The agent MUST verify that leading indicators (support tickets, usage decline, stakeholder changes) are weighted appropriately
- The agent MUST verify that risk classifications trigger specific action plans, not just color codes
- The agent MUST verify that no at-risk accounts are classified as healthy due to recency bias
Health Check
Criteria Guidance
Good criteria examples:
- "Health scorecard rates account across at least 5 dimensions (usage, engagement, support sentiment, stakeholder access, contract alignment) with evidence for each rating"
- "Risk assessment identifies specific churn indicators with severity levels and mitigation timelines"
- "Action plan includes owner assignments and measurable success criteria for each remediation item"
Bad criteria examples:
- "Account health is assessed"
- "Risks are identified"
- "Action plan exists"
Completion Signal (RFC 2119)
Health report MUST exist with a scored health assessment across multiple dimensions. Risk analyst MUST have identified and severity-ranked all churn risks with specific leading indicators. Each risk MUST have a concrete mitigation plan with success criteria. Escalation paths MUST be documented for critical risks. Report provides a clear recommendation: healthy, at-risk, or critical.