Skip to content

Quick Start

See goflow in action in under 2 minutes. We'll run an existing example workflow.

Step 1: Get goflow

If you haven't installed goflow yet, do it now:

git clone https://github.com/alxayo/goflow.git
cd goflow
go build -o goflow ./cmd/workflow-runner/main.go

Step 2: Run the Example Workflow (Mock Mode)

The repository includes example workflows. Let's run one in mock mode, which simulates AI responses:

./goflow run \
  --workflow examples/simple-sequential.yaml \
  --inputs files='pkg/workflow/*.go' \
  --mock \
  --verbose

What you'll see

[INFO] Loading workflow: examples/simple-sequential.yaml
[INFO] Starting workflow: simple-sequential
[INFO] Step 1/3: security-review (security-reviewer)
[INFO] Step 2/3: perf-review (performance-reviewer)  
[INFO] Step 3/3: summary (aggregator)
[INFO] Workflow completed in 0.05s

## Step: summary

mock output

Why Mock Mode?

  • No API calls — doesn't use real AI tokens
  • Instant results — runs in milliseconds
  • Full pipeline test — validates your YAML structure, dependencies, and templates

Use mock mode while developing workflows, then switch to real mode when ready.

Step 3: Inspect the Audit Trail

Every run creates a complete record in .workflow-runs/:

ls .workflow-runs/

You'll see a timestamped folder like 2026-03-26T10-15-30_simple-sequential/.

Look inside:

ls .workflow-runs/*/

workflow.meta.json    # Run metadata and timing
workflow.yaml         # Snapshot of the workflow
final_output.md       # The formatted output
steps/                # Individual step data

Each step has its own folder with the exact prompt sent and response received:

cat .workflow-runs/*/steps/00_security-review/prompt.md
cat .workflow-runs/*/steps/00_security-review/output.md

Step 4: Run With Real AI (Optional)

If you have Copilot CLI installed, try a real run:

./goflow run \
  --workflow examples/simple-sequential.yaml \
  --inputs files='pkg/workflow/*.go' \
  --verbose

Remove --mock and goflow will make actual AI calls. The output will be real code review content instead of "mock output".

What Just Happened?

The simple-sequential.yaml workflow:

  1. Defined 3 agents — security reviewer, performance reviewer, and aggregator
  2. Ran 3 steps in sequence:
  3. security-review — reviewed files for security issues
  4. perf-review — reviewed files for performance issues (waited for security-review)
  5. summary — combined both reviews (waited for perf-review)
  6. Passed data between steps — using {{steps.security-review.output}} templates
  7. Created an audit trail — saved everything to .workflow-runs/

Next Steps