Report missing safe outputs in failure issue instead of failing conclusion job

[Copilot]

Changes Addressing Feedback

Updated approach per @pelikhan's feedback: Instead of failing the conclusion job when no safe outputs exist, the information is now included in the failure tracking issue/comment.

Changes Made:

• Modified handle_agent_failure.cjs to detect when agent succeeds but produces no safe outputs
• Added missing safe outputs context to failure issue/comment templates
• Removed the "Determine if conclusion job should fail" step from conclusion job
• Updated tests to cover new behavior (agent success with/without safe outputs)
• Recompiled all workflows
• Removed unused determine_conclusion_failure.cjs and test file

Behavior:

When agent succeeds but produces no safe outputs:

• Creates/updates failure tracking issue with warning about missing outputs
• Does NOT fail the conclusion job
• Explains possible causes (safe output server failure, prompt failure, missing noop call)

When agent succeeds with safe outputs:

• Skips failure handling (normal behavior)
• Conclusion job completes successfully

Original prompt

Update the JavaScript that decide to report a failure in the conclusion job to report workflow run without any safe output entries as failures. Typically this means that the safe output server failed to run or that the prompt completely failed to generate any kind of meaningful result and it should have called "noop".

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[pelikhan]

@copilot do not fail the conclusion, instead add this information in the action failed issue/comment created in the conclusion job.

[Copilot]

@copilot do not fail the conclusion, instead add this information in the action failed issue/comment created in the conclusion job.

Updated in commit 7babd57. Instead of failing the conclusion job, the missing safe outputs information is now included in the failure tracking issue/comment created by handle_agent_failure.cjs.

The behavior now:

• When agent succeeds but produces no safe outputs → creates/updates failure issue with warning
• When agent succeeds with safe outputs → skips failure handling (normal flow)
• Conclusion job never fails due to missing safe outputs