Skip to content

Recipes

Goal-oriented patterns for the operations plugin authors do most often. Each recipe is short — a goal, a code block, a few notes, and links to related material. If the recipe doesn't fit, the Tutorials cover end-to-end builds and the API Reference covers exact symbol surfaces.

Reading experiments

RecipeWhen
Reading experimentsLook up one experiment, paginate, filter by status / project
Querying plugin dataSQLAlchemy patterns on plugin-owned tables

Writing results

RecipeWhen
Writing resultsSave / append PluginAnalysisResult, preserve run history, reference produced artifacts by ID

Permissions and identity

RecipeWhen
Route permissionsrequire_plugin_role, combining with platform permissions

Reliability

RecipeWhen
Error handlingMap PluginException subclasses to HTTP responses; user-facing vs developer-facing
Logging & tracingStructured logs via get_plugin_logger; request-scoped fields
Testing pluginsIn-memory repos, fixtures, end-to-end tests with TestClient

Schema evolution

RecipeWhen
Backfill migrationsSchema changes plus chunked data backfill, idempotent on retry

Domain-specific

RecipeWhen
R integrationCalling R from a Python plugin via rpy2 or subprocess

Pattern

Every recipe page follows the same shape: Goal → Code → Notes → Related. Read the goal and the code first; the notes are the gotchas you'll hit when adapting the pattern.

MINT is open source. Made by the Morscher Lab.