Patterns for adding extractors and write surfaces to the PR 12 fact-extraction pipeline (`apps/erify_api/src/orchestration/fact-extraction/`). Use BEFORE implementing any new `IngestionExtractor`, paired-atomic write, `SystemFactKey`, or hydrated-scope target type. Required reading before PR 12.3.2 (violations), 12.4 (review surface), or any follow-on extractor work — every Codex finding on PR 12.1.2 (#103) and PR 12.2 (#104) is encoded here. The "State-transition handoff between co-submitted facts" section is mandatory before adding ANY fact whose write semantics depend on another fact's value in the same submission.
Patterns for auditing and improving erify_api query performance and response efficiency. Use when detecting N+1 queries, reducing over-fetching, designing lean select/include strategies, replacing in-memory joins with DB aggregations, adding pagination guards, or defining API performance baselines before scaling. Complements database-patterns which covers the basic N+1 and Promise.all rules.
Patterns for implementing and extending the Studio Shift Schedule feature. This skill should be used when working on shift CRUD, shift blocks, calendar/alignment orchestration, duty-manager coverage, operational-day boundaries, task-readiness checks, or the frontend shift schedule UX.
Provides guidelines for the Task Template Builder architecture, including Schema alignment, Draft storage, Drag-and-Drop, and Validation logic. Use when implementing or modifying the task template builder UI, draft persistence, or template validation flows.
Provides Prisma-specific patterns for soft delete, transactions, optimistic locking, advisory locks, bulk operations, and performance optimization. This skill should be used when implementing data persistence, handling concurrent updates, serializing concurrent operations on the same resource, managing complex multi-table operations, or optimizing query performance.
Use when expanding the SystemFactKey catalog, onboarding task templates that need operational fact extraction, a prod-db-sync reveals task templates without system_fact_key bindings, or a field-type mismatch (e.g. select where checkbox or multiselect is required) blocks binding a template field to the fact-extraction pipeline.
Provides patterns for building and optimizing large frontend tabular views across the eridu-services monorepo. Use when building or refactoring server-driven tables, dense tabular UIs, virtualized grids, inline editing, saved views, and table URL state in erify_creators, erify_studios, or shared @eridu/ui.
Provides code quality standards for frontend applications. This skill should be used when configuring linting rules, organizing file structures, or ensuring consistency across React applications.