Improve CI scanner duplicate KBE detection

[svick]

Why

The CI failure scanner filed #130940 even though #129691 already covered the same failure. The run searched a shortened underscore-delimited method name, which GitHub did not match as a prefix of the complete test identifier.

The scanner instructions also claimed that Build Analysis results were available through an AzDO attachment named Build_Analysis_KnownIssues_v1. Investigation found no producer or observed instance of that attachment. Build Analysis did recognize the failure, but that information was exposed through its GitHub check instead.

Changes

• Search complete test method identifiers verbatim before deriving shorter family stems.
• Replace the nonexistent AzDO attachment with a best-effort lookup of the Build Analysis GitHub check.
• Treat links found in the check as positive dedup evidence, while continuing exact KBE searches after a miss because large check reports are truncated.

Validation

A targeted dry-mode eval replayed build 1505119. The rendered Build Analysis check omitted #129691, then the verbatim full-method search found it and produced existing-kbe #129691 with no issue creation. All 8 targeted graders passed.

cc: @kotlarmilos

Note

This pull request description was generated with GitHub Copilot.

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).
15 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/runtime-infrastructure
See info in area-owners.md if you want to be subscribed.

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).
15 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[Copilot]

Pull request overview

Updates the CI Outer-Loop Failure Scanner guidance to reduce duplicate Known Build Error (KBE) filings by improving how it searches for existing KBEs and by leveraging Build Analysis results exposed via a GitHub check instead of an AzDO attachment.

Changes:

• Add guidance to search full test method identifiers verbatim (especially underscore-delimited names) before deriving shorter stems.
• Update the scanner’s “Build Analysis” lookup instructions to use the GitHub check-runs API (instead of a referenced-but-nonexistent AzDO attachment).
• Refresh the generated workflow lock metadata hash to reflect the updated source workflow text.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
.github/workflows/shared/create-kbe.instructions.md	Adds guidance to avoid truncating underscore-delimited test identifiers during existing-KBE searches.
.github/workflows/ci-failure-scan.md	Replaces Build Analysis “attachment” guidance with a best-effort lookup via the Build Analysis GitHub check.
.github/workflows/ci-failure-scan.lock.yml	Updates gh-aw metadata (body_hash) to match the edited workflow source.

[kotlarmilos]

One minor comment, otherwise LGTM!

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)

.github/workflows/shared/create-kbe.instructions.md:68

• The semicolon after "underscore-delimited identifiers" reads like an accidental sentence break; using a period here improves grammar and avoids implying an additional clause.

When a failure includes a complete test method identifier, search that
identifier verbatim before deriving any shorter stem. Do not truncate
underscore-delimited identifiers;
GitHub search does not reliably prefix-match them. Only strip the specific
platform, architecture, and type-width suffixes described in variation 6.