Skip to content

Getting Started

This page gives you a structured path from installation to building production-ready flows.

1. Install

Install with pip install auto-workflow or see Quickstart for development setup.

2. Learn Core Primitives

  1. Read Tasks
  2. Read Flows
  3. Skim Dynamic Fan-Out for mapping patterns

3. Build Your First Flow

Follow the minimal example in Quickstart. Run it and inspect with describe() / export_dot().

4. Add Reliability

Configure retries, backoff, and timeouts: Retries & Failures

5. Optimize Execution

  • Use priority to schedule important tasks earlier
  • Apply max_concurrency if your environment has resource limits

6. Manage Data

Decide per task: - Large payload? -> persist=True (see Caching & Artifacts) - Expensive but deterministic? -> cache_ttl=...

7. Observe & Debug

  • Subscribe to events (see Middleware & Events)
  • Add logging middleware for structured logs
  • Export a DOT graph for visualization

8. Handle Secrets

Integrate your secrets provider per Secrets before embedding credentials.

9. Extend

Need custom caching, tracing, or metrics? See Extensibility.

10. Prepare for Production

Checklist: - [ ] All tasks idempotent or guarded by cache - [ ] Retries set on transient operations - [ ] Timeouts defined for network-bound tasks - [ ] Artifact persistence for large objects - [ ] Events subscribed for alerts / metrics exported - [ ] Flow graph exported & documented

  • Use the left sidebar (MkDocs) to jump sections.
  • Each page has an auto-generated table of contents (right side if theme supports) due to toc extension.
  • Start broad (Quickstart) then dive deeper (API Reference) as needed.

Common Next Questions

Goal Where to Look
"How do I dynamically map tasks?" Dynamic Fan-Out
"How do I skip recomputation?" Caching & Artifacts
"How do I trace performance?" Observability
"Where is the public API list?" API Reference

If anything feels missing, open an issue or PR—this page should remain the fastest on-ramp.