Route Work with Structured Agent Outputs

Use structured outputs when later workflow logic needs a value from an agent. This guide records a summary and a boolean review decision instead of parsing the agent's final response.

Before you begin

You need a project that can sync workflows, an online runner with the ubuntu-latest label, and configured agent defaults.

Add the workflow

Create .shipfox/workflows/classify-release-risk.yml. This is a complete workflow:

# yaml-language-server: $schema=https://www.shipfox.io/docs/workflow.schema.json
name: Classify release risk
runner: ubuntu-latest

triggers:
  manual:
    source: manual
    event: fire

jobs:
  assess:
    steps:
      - key: assess
        prompt: |
          Review the repository for release risk.
          Set summary to one sentence that explains the main risk.
          Set needs_review to true when a person should review before release.
        outputs:
          summary: string
          needs_review: boolean

      - env:
          SUMMARY: "${{ steps.assess.outputs.summary }}"
        run: |
          printf 'Assessment: %s\n' "$SUMMARY"

      - if: '${{ steps.assess.outputs.needs_review == true }}'
        run: echo "Human review required"

The declared outputs give the agent two set_output tools. The summary is bound through env before shell use. The final step reads the boolean directly in its condition.

Commit and push the file to the project's default branch. Wait for Classify release risk to sync.

Verify the decision

Start the workflow and open the assess job.

1. Confirm that the agent sets both declared outputs.
2. Confirm that the assessment line matches the summary output.
3. If needs_review is true, confirm that the final step runs. If it is false, confirm that Shipfox marks the final step skipped.

An agent step fails when it does not set a required output after its bounded re-prompts. See Step outputs for the exact behavior.