Instruction file imported from hazeydata/theme-park-crowd-report-2 (
.cursor/rules/keep-pipeline-state.mdc). Copyright stays with the author.
Pipeline state doc
docs/PIPELINE_STATE.md is the single "where we are" reference (config paths, cron, queue-times, dashboard, commands).
When you change any of the following, update docs/PIPELINE_STATE.md so it stays accurate:
- output_base or default paths (config examples,
src/utils/paths.py, config README) - Cron (what runs when,
scripts/install_cron.sh,run_daily_pipeline.sh, LINUX_CRON_SETUP) - Queue-times (how it runs, systemd service,
install_queue_times_service.sh,queue-times-loop.service) - Dashboard (how to run it, what it reads,
dashboard/app.pyor README) - Key paths (state/, logs/, aggregates/, etc. under output_base)
- Commands (manual daily run, queue-times manual/systemd, prerequisites)
If you only change application logic (e.g. parsers, training, ETL logic) and not where things run or how to run them, you don't need to update PIPELINE_STATE.
When in doubt, skim PIPELINE_STATE and adjust the summary table, paths, or command blocks to match the current setup.