Large collection loaded then processed in memory

How to fix

1. Push filtering, projection, and aggregation into the query.
2. Apply limits before loading records.
3. Move large exports to explicit batch workflows.

What it means

A workflow appears to fetch more rows than it needs before applying filtering or aggregation.

Why it matters

In-memory processing can degrade latency and memory use for large tenants or repositories.

Common causes

• Filtering was easier to express in application code.
• Pagination was applied after fetching all rows.
• A report path reused online request code.

Example bad pattern

findMany(...) loads all findings, then filters open findings in JavaScript.

Example good pattern

findMany({ where: { status: 'open' }, take: pageSize }).

Related files/config

• apps/api/src
• packages/db/prisma/schema.prisma

Related CLI commands

• archpilot validate
• archpilot validate --ci