# Planning document lifecycle example

OpenAI Codex as Directed by Tyson Weihs.

This is executable instructional code using the existing Rust producer and independent
Python reader. All aircraft, weather, navigation and elevation inputs in `request.json`
are invented specimens. Airport identifiers provide recognizable route context, not
operational data. No native handles, local paths, customer plans or licensed records
are included in the downloadable input.

From the Aeronautiql source checkout, with Rust and Python 3.11+ available:

```sh
cargo build --locked --manifest-path modules/planning/rust/planning_engine/Cargo.toml
python3 -m venv /tmp/aeronautic-example-venv
/tmp/aeronautic-example-venv/bin/pip install jsonschema==4.26.0
/tmp/aeronautic-example-venv/bin/python modules/planning/examples/document-lifecycle/run.py --output /tmp/aeronautic-example
```

The script needs its source checkout: it imports owner validators and locates the native
binary relative to this directory. Downloading this script alone does not install an SDK.
It emits intent, captured request, result, snapshot and draw JSON into the chosen directory.
Assertions verify the happy path and stale/invalid/unavailable cases. A failure exits nonzero.
The sample does not persist an app database, reconcile cloud conflicts or acquire live weather;
those host workflows are covered by the Planning coordinator and PlanningNative tests.

Do not compute dependency hashes or resolve map hits independently in a client. Native
`capture-resources`, `prepare-key`, `edit-mapping` and `map-edit` own those operations.
An HTTP service may wrap this dispatcher, but this example does not advertise a deployed
public HTTP planning endpoint.
