Build an Agent Feedback Loop in Shipfox

This guide builds a workflow where an agent fixes failing tests and npm test decides whether another attempt is needed.

Before you begin

You need a test project that can sync workflows, an online runner with the ubuntu-latest label, configured agent defaults, and an npm test command with a safe failure that the agent can repair.

Add the workflow

Create .shipfox/workflows/fix-tests.yml. This is a complete workflow:

# yaml-language-server: $schema=https://www.shipfox.io/docs/workflow.schema.json
name: Fix failing tests
runner: ubuntu-latest

triggers:
  manual:
    source: manual
    event: fire

jobs:
  fix:
    execution_timeout: "30m"
    steps:
      - key: fix
        prompt: >
          Fix the failing tests in this repository.
          Previous check: ${{ step.is_retry
            ? step.restart.feedback
            : "No previous check. Run the tests and inspect the failures." }}

      - key: test
        run: npm test
        gate:
          success: step.exit_code == 0
          on_failure:
            restart_from: fix
            feedback: The tests still fail. Run them again and fix the cause.

Replace npm test with the command that proves the work is correct in your repository.

Run and inspect it

1. Commit and push the workflow to the project's default branch.
2. Wait for its definition to sync.
3. Open the workflow and select Run.
4. Open the fix job in the run detail.
5. If the check fails, confirm that a new fix step attempt receives the gate feedback.
6. Confirm that the job finishes when the check passes or the restart bound is exhausted.

The key on fix is required because restart_from targets step keys. The target must appear before the gated step in the same job. Exact gate fields and validation rules live in Gate fields.

The job timeout is an outer wall-clock bound. The gate also has its own attempt limit. See Limits. For the evaluation model, see Feedback loops.