# Wetech Explicit Cluster Clarification

<!-- auto-oncall-optimization-record
{
  "schemaVersion": "auto-oncall-optimization-record-v1",
  "optimizationId": "wetech-explicit-cluster-clarification-2026-08-21",
  "problems": [
    {
      "problemId": "wetech-router-guessed-regional-saas-default-without-ticket-target",
      "summary": "A Wetech diagnostic ticket that named no cluster was routed to asia-prod / SaaS01 instead of asking the customer to identify the affected cluster.",
      "status": "optimized_unverified",
      "applicability": "Wetech and apac-test Provider-backed tickets with no explicit customer-supplied cluster alias, business name, or full cluster name.",
      "issueKeys": ["APID-29369"],
      "sourceRefs": ["https://tidb.atlassian.net/browse/APID-29369#problem:missing-cluster-target", "https://tidb.atlassian.net/browse/APID-29369#problem:unsupported-saas01-assumption"],
      "signalTextIncludesAny": ["未指明是哪个集群", "直接认为是 saas01 集群缺乏依据", "no explicit cluster target"],
      "observedThrough": "2026-08-21",
      "effectiveAt": "2026-08-21T00:00:00Z",
      "verifiedAt": null,
      "attemptedAction": "Require an explicit customer-supplied Wetech cluster target at classification and Scenario Router assignment; emit cluster_alias clarification when absent and remove the regional SaaS default.",
      "evidenceUrl": "https://github.com/tidbcloud/aisupport/blob/main/auto-oncall-services/docs/opts/2026-08-21-wetech-explicit-cluster-clarification.md"
    }
  ]
}
-->

- Owner: Scenario Classifier missing-context decision and Wetech Scenario Router target assignment
- Source: live Jira `APID-29369`, comments `23625097` and `23626426`
- Implementation status: local candidate
- Effectiveness status: `optimized_unverified`

## Bounded problem

The ticket identified the deployment type and Grafana symptom but supplied no
cluster name. The generated internal handoff nevertheless selected
`asia-prod / SaaS01`. The earliest broken boundary was the Router rule that
converted an absent Wetech target into a regional SaaS default, reinforced by
classification treating the Wetech Account Alias as sufficient cluster context.

## Candidate change

- Treat the Account Alias as Provider selection only, not cluster identity.
- When no explicit Wetech cluster alias, business name, or full name is present,
  classify the request as `clarification_needed` with
  `missingKeys=[cluster_alias]` and ask for the affected cluster name and
  available cluster-identifying information.
- Make the Scenario Router return no entity when the ticket has no explicit
  target. Do not infer `SaaS01`, `SaaS`, or another cluster from region, Jira
  key, account, or another ticket.
- Preserve internal unique resolution after the customer supplies a target;
  an internal resolution miss remains Support-owned and does not trigger a
  duplicate customer question.

## Verification evidence

The focused classifier and Router contract tests cover the original missing-
target shape and the removal of regional default selection. Deployment and a
fresh Provider-backed replay are intentionally outside this change; the record
remains `optimized_unverified` until runtime rollout evidence exists.

## Unresolved gaps

- No deployment was performed, so staging behavior has not changed yet.
- No fresh runtime replay was available to validate the final paired response.

## Next validation action

After rollout, replay `APID-29369` at the original cutoff and verify that the
customer draft asks for the affected cluster name and identifying information,
the internal handoff contains no assumed target, and no Wetech query starts
before the clarification is supplied.