# Expected Diagnosis: O11Y-2634149

> Note: This case was sourced from user feedback (GitHub issue #992). The workspace
> (lark-ws-1d2debe1-f617-44f2-856e-f030abf122ee) is in production and not accessible
> for regression replay. This expected.md documents the known failure mode only.

## Root Cause

The O11Y alert contained a cluster ID that belongs to a k8s infrastructure cluster
(the cluster running the observability/monitoring system), not the actual customer TiDB
cluster with the performance issue. The agent must detect this mismatch and ask the user
for the correct TiDB cluster ID before proceeding with diagnosis.

## Evidence Chain

1. O11Y alert's cluster ID → Clinic cluster metadata lookup → cluster_type indicates infra/ops cluster
2. User confirms: cluster ID = k8s infra cluster, not the affected TiDB cluster
3. Correct action: halt diagnosis, report the issue to user, request correct cluster ID

## Causality

O11Y alert cluster ID references k8s infra cluster → Agent runs Clinic evidence collection on wrong cluster → Diagnosis conclusions are irrelevant to actual problem cluster

## Severity

High — diagnosis is completely misdirected; user feedback confirms no useful output.

## Recommendations

1. Validate cluster ID before starting evidence collection: check cluster metadata for signs of infra/ops cluster
2. If validation fails or is uncertain, prompt user to confirm the cluster ID is correct
3. Document known infra cluster ID patterns in SKILL.md to help agent identify them

## Scoring Rubric

> Checklist for regression testing. The Agent must satisfy these to pass.

- [ ] Agent detects that the cluster ID in the O11Y alert belongs to a k8s infrastructure cluster (not a customer TiDB cluster)
- [ ] Agent does NOT proceed with Clinic data collection on the k8s infra cluster
- [ ] Agent explicitly informs the user that the cluster ID appears to be a k8s infra cluster
- [ ] Agent asks the user to provide the correct TiDB cluster ID for diagnosis