Skip to content

CLAUDE.md

Latest commit

 

History

History
53 lines (36 loc) · 1.64 KB

CLAUDE.md

File metadata and controls

53 lines (36 loc) · 1.64 KB
Raw

Claude Code Instructions

Persistence

When asked to remember something or when learning reusable guidelines, save them in this file — not in local memory. This repo is used across multiple machines, so only repo-level persistence is reliable.

Before committing

Always run pre-commit run --files <staged files> before committing to catch formatting and linting issues. Fix any failures before creating the commit.

After modifying SQL queries

Validate every changed SQL file against BigQuery with --dry_run before committing:

bq query --project_id=idc-sandbox-000 --use_legacy_sql=false --dry_run < path/to/query.sql

Fix any query errors before creating the commit. Note that prior_versions_index.sql uses procedural SQL (EXECUTE IMMEDIATE) and cannot be validated with --dry_run.

Before starting any task

Read the developer documentation in docs/dev/ before exploring the codebase. In particular, check if there is a guide that matches your task (e.g., docs/dev/adding-new-index.md when adding a new index). Use the guide as your checklist.

BigQuery access

Queries can be run locally via:

bq query --project_id=idc-sandbox-000 --use_legacy_sql=false < path/to/query.sql

Column naming convention

  • DICOM attributes use PascalCase (e.g., SeriesInstanceUID, PatientID)
  • Derived/computed columns use snake_case (e.g., sop_class_name, series_size_MB, valid_3d_volume)

Index SQL locations

  • Core indexes: scripts/sql/ (idc_index, collections_index, prior_versions_index, analysis_results_index)
  • Specialized indexes: assets/ (clinical_index, sm_index, sm_instance_index, seg_index, etc.)