LEX Docs Topic Map for LEX AI Users

How the shipped docs/lex_topics reference is organised, how to retrieve from it efficiently, and why the shipped topics override a model's prior knowledge.

A
Written By Atakan KayaLast updated 3 months ago

LEX Docs Topic Map for LEX AI Users

The docs/lex_topics collection is a focused reference map for Lex framework questions. LEX AI installs it into your project so the assistant answers from the shipped handbook rather than from whatever it happens to remember about Django.

Topic coverage model

  • Architecture and runtime
  • Project structure and discovery
  • LexModel core, CalculationModel lifecycle, and CalculatedModelMixin combinatorics
  • Lifecycle hooks
  • Permissions and authorization; authentication and Keycloak
  • Serializers and the API layer; fields and report assets
  • Streamlit dashboards
  • ProcessAdmin registration and model_structure.yaml
  • Logging, Celery async dispatch, transactions and deferred recalculation, signals and websocket updates
  • Initial data upload; the lex CLI, settings, imports and utilities
  • Testing and test data
  • Patterns, pitfalls and checklists; examples and endpoints
  • Canonical Lex specifications and the baked framework context file
  • A query router for fast file selection

Retrieval guidance

  • Start with one best-match topic file. 99-QUERY-ROUTER.md maps keywords to the right one and 00-TOPIC-LIST.md is the index.
  • Open a second file only when needed.
  • Expand to broader context files only when key details remain missing.

Typical use

If a user asks a narrowly scoped framework question, route to the matching topic file instead of reading large handbook sections first. Two files carry unusual weight during a LEX AI run: 23-testing-and-test-data.md, because in Lex the JSON action chain replayed by ProcessAdminTestCase is the test rather than merely its data, and 17-cli-settings-imports-utils.md, because that is where installing and running an app is documented — there is no separate getting-started page.

Why the handbook ships with the product

When the shipped topics contradict a model's prior knowledge, the topics win. History and audit logging, for instance, are automatic Lex features that a Django-trained model will try to implement by hand. Installing the topics locally is what makes that rule enforceable rather than aspirational.

Was this helpful?

Your feedback shapes what we write next.