Description of problem:

On Forgejo, when a user runs /retest on a pull request where the Pipelines as Code CI has already succeeded (no failures), the system posts a "Pipelines as Code CI has failed" comment and adds a failure check/logo. The pipeline run itself remains successful, so the PR shows both a success check and a failure check. The failure is incorrect and misleading.

The failure comment states: "could not find any PipelineRun in your .tekton/ directory", suggesting that the /retest flow re-validates the commit in a way that does not find the existing pipeline definition (e.g. remote annotations or non-.tekton location), and then reports that as a CI failure even when there were no actual test failures.

Prerequisites (if any, like setup, operators/versions):

• Forgejo (e.g. 13.0.4+gitea-1.22.0 or similar)
• Pipelines as Code (Pac) configured with Forgejo
• A repository using Pipelines as Code (e.g. pipeline defined via remote annotations or other mechanism, not necessarily under .tekton/)
• A PR where Pac has already run and reported success

Steps to Reproduce

1. Open a PR that uses Pipelines as Code (e.g. remote annotations).
2. Wait for Pac to run and post a success comment (e.g. "Pipelines as Code CI/piplinerun-... has successfully validated your commit").
3. In the PR conversation, comment /retest.
4. Observe the new Pac comment and the PR checks section.

Actual results:

• Pac posts a comment: "Pipelines as Code CI has failed" with "could not find any PipelineRun in your .tekton/ directory".
• The PR checks show both:
  • Pipelines as Code CI / piplinerun-... — Success
  • Pipelines as Code CI — Failed
• A failure logo/status is shown even though the pipeline run had no failures and the PR was mergeable.

Expected results:

• When /retest is used and there are no actual pipeline failures, no additional failure comment or failure check should be posted.
• The PR should not show a failure logo for Pipelines as Code CI when the pipeline has succeeded; at most, the existing success status should remain or be refreshed without adding a spurious failure.

Reproducibility (Always/Intermittent/Only Once):

Always — reproducible when /retest is used after a successful run on a repo where pipeline is not defined under .tekton/ (e.g. remote annotations).

Acceptance criteria:

• When /retest is run and the pipeline has no failures, no "Pipelines as Code CI has failed" comment is posted.
• No failure check/logo is added for Pipelines as Code CI when the pipeline actually succeeded.
• Behavior is correct for both .tekton/-based and remote-annotation (or other) pipeline definitions.
• Existing success reporting remains correct.

Definition of Done:

• Root cause of /retest re-validation (e.g. .tekton/ lookup) is identified and fixed or bypassed when appropriate.
• Fix verified on Forgejo with a successful pipeline and /retest.
• No spurious failure comments or checks when there are no failures.

Additional info (e.g. logs, screenshots):

Screenshot available showing: successful "Pipelines as Code CI/piplinerun-remote-annotations-rqd5q" run, then /retest comment, then "Pipelines as Code CI has failed" with "could not find any PipelineRun in your .tekton/ directory", and the PR checks showing both Success and Failed for Pipelines as Code CI.