AI-Assisted Coding With IoP

Use this guide when you want to vibe code with IoP while still getting production-shaped, reviewable Python interoperability code.

Start Here

Ask the tool to read the project guidance before it writes code:

You are helping me build an IoP application.
Read these files first:
- AGENTS.md
- docs/ai-coding.md
- docs/cookbooks/index.md
- docs/getting-started/first-steps.md
- docs/getting-started/register-component.md
- docs/production-graph.md

Use the Python Production graph workflow for new applications.
Do not put component startup logic in __init__(); use on_init().
Use the relevant cookbook for the task.
Show the migration and verification commands.

On the published docs site, open AGENTS.md from the repository root.

For your own IoP application repository, copy the reusable AGENTS.md template into the project root and adapt it to the local production. For healthcare projects, also read Healthcare AI-assisted coding.

Cookbooks

Use the IoP cookbooks for task-specific prompts and checklists:

Good Output Expectations

For a new IoP application, generated code should include:

  • a settings.py file with a Production object
  • PRODUCTIONS = [prod]
  • clear component names such as FileService, RouteProcess, or OrderOperation
  • message classes for data exchanged between components
  • target() settings for configurable outbound routing
  • prod.connect(...) calls that wire services, processes, and operations
  • sample payloads or tests when behavior changes
  • migration and verification commands

Useful Commands

# Fast pure-Python checks
python -m pytest src/tests/unit

# Validate a migration file without writing to IRIS
iop --migrate settings.py --dry-run

# Migrate a production
iop --migrate settings.py

# Build the documentation
mkdocs build

# Run the Docker-backed suite
docker build -t pytest-iris -f dockerfile-ci .
docker run -i --rm pytest-iris