Adding fix workflow to failing pipelines

[ReilleyMilne (ReilleyMilne)]

• Simplifies the analysis workflow.
• Adds an additional agentic workflow to automatically suggest fixes to failing pipelines. This extends the already existing analysis workflow.
• Currently the fix is done in an agent authored branch + PR, so this PR also includes a workflow to clean up those opened branches. The PR's opened by GitHub agentic workflows are automatically closed by the agentics-maintenance.yml run.

Example Demos:
Analysis: ReilleyMilne#58
Fix: ReilleyMilne#59

Fixes: Azure/azure-sdk-tools#16205

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).
9 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

Adds automated diagnosis and draft fix proposals for failing Azure SDK pipelines, addressing Azure/azure-sdk-tools#16205.

Changes:

• Simplifies pipeline analysis with direct check_suite handling.
• Adds an agentic auto-fix workflow that creates draft fix PRs.
• Adds cleanup for completed auto-fix branches.

Reviewed changes

Copilot reviewed 5 out of 6 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
.github/workflows/pipeline-analysis-next-steps.md	Updates analysis and auto-fix dispatch logic.
.github/workflows/pipeline-analysis-next-steps.lock.yml	Compiles the updated analysis workflow.
.github/workflows/pipeline-analysis-next-steps-trigger.yml	Removes the superseded dispatcher.
.github/workflows/pipeline-analysis-branch-cleanup.yml	Cleans obsolete auto-fix branches.
.github/workflows/pipeline-analysis-auto-fix.md	Defines draft fix generation and retargeting.
.github/workflows/pipeline-analysis-auto-fix.lock.yml	Compiles the auto-fix workflow.

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 6 changed files in this pull request and generated no new comments.

Suppressed comments (2)

.github/workflows/pipeline-analysis-next-steps.md:180

• failed_pipeline_tests is a platform-to-test-title map, not a collection containing artifact_file_path (the checked-in pipeline-analysis skill likewise describes title lists). Therefore this condition can never obtain a path, and because azsdk_get_pipeline_llm_artifacts is absent from the MCP allowlist, the workflow cannot fetch the detailed test data that the following sentence requires before diagnosis/fixability classification. Allow and call azsdk_get_pipeline_llm_artifacts first, then pass its returned artifact paths to the test-run tools.

4. Inspect every `failed_pipeline_tests` entry returned by the analysis. If an
  `artifact_file_path` is present, call `azsdk_get_failed_test_run_data` exactly once per unique
  artifact with `failedTestRunsPath` set to that path. Call `azsdk_get_failed_test_case_data` only

.github/workflows/pipeline-analysis-next-steps.md:43

• The replaced trigger explicitly skipped draft PRs, but this new gate treats drafts as eligible. As a result, expected failures while an author is still drafting can consume Copilot runs and automatically open fix PRs. Preserve the previous draft guard unless enabling automation on drafts is intentional.

          const matchingPulls = pulls.filter(
            pull => pull.state === "open" && pull.head.sha === suite.head_sha
          );

[Ben Broderick Phillips (benbp)]

It would be nice if we can work in some logging in a pre/post step around the main work to show the impact on github API rate limits. We have an azure pipelines example here: https://github.com/Azure/azure-rest-api-specs/blob/5b13925393aba9cb9248171d1bbab9287336f569/eng/common/pipelines/templates/steps/emit-rate-limit-metrics.yml#L14 but Mike Harder (@mikeharder) may be able to show you a better example of where we are doing it in github workflow contexts.

[Ben Broderick Phillips (benbp)]

Otherwise, LGTM