Report agent assignment failures without blocking safe_outputs job

[Copilot]

Agent assignment failures currently fail the safe_outputs job via core.setFailed(), blocking other safe output operations (issue creation, comments, etc). Assignment errors should be reported in the conclusion job without failing safe_outputs.

Changes

actions/setup/js/assign_to_agent.cjs

• Replace core.setFailed() with core.warning() when assignments fail (line 431)
• Error details remain captured in job outputs (assignment_errors, assignment_error_count)
• Configuration validation errors still fail the job as expected

actions/setup/js/assign_to_agent.test.cjs

• Update 7 test assertions: expect(mockCore.setFailed) → expect(mockCore.warning)

Flow

// Before: Job fails, blocks other safe outputs
if (failureCount > 0) {
  core.setFailed(`Failed to assign ${failureCount} agent(s)`);
}

// After: Job continues, errors reported in conclusion
if (failureCount > 0) {
  core.warning(`Failed to assign ${failureCount} agent(s) - errors will be reported in conclusion job`);
}

The handle_agent_failure.cjs script already reads these outputs from safe_outputs job and formats them in failure issues/comments (lines 339-340, 429-444, 511-527).

Original prompt

Reference: https://github.com/github/gh-aw/actions/runs/21672174594/job/62483376429#step:7:1

Failure to assign to agent must be reported in the agent failure in the conclusion job in the issue/comment. Do not fail the safe outputs job.

See attached report

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[Copilot]

Pull request overview

This PR ensures that agent assignment failures no longer cause the safe_outputs job to fail, allowing downstream safe-output operations (issues, comments, etc.) to still run while surfacing assignment problems via the conclusion job.

Changes:

• Update assign_to_agent.cjs to replace core.setFailed with core.warning when one or more agent assignments fail, while still setting assignment_errors and assignment_error_count outputs for the conclusion job.
• Keep configuration validation errors (e.g., invalid GH_AW_AGENT_MAX_COUNT) failing the job via core.setFailed, preserving strictness for misconfiguration.
• Adjust unit tests in assign_to_agent.test.cjs to expect warnings instead of job failure for assignment error paths.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
actions/setup/js/assign_to_agent.cjs	Switches agent-assignment failure handling from failing the job to emitting a warning, while preserving outputs consumed by handle_agent_failure.cjs.
actions/setup/js/assign_to_agent.test.cjs	Updates tests to assert on core.warning for assignment failures and maintains existing checks around core.setFailed only for configuration/other non-assignment paths.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Given the PR’s goal is to stop failing the safe_outputs job on agent-assignment errors, this test now only asserts that a warning is logged but no longer checks that core.setFailed is not called. To lock in the new behavior and prevent regressions where assignment failures start failing the job again, consider adding an assertion like expect(mockCore.setFailed).not.toHaveBeenCalled() here (and/or in at least one other failure-path test).