Plain-language steps to rebuild every output from the client's files. Run each command from the project folder (the one that holds data/ and my-work/), in this order. Each step names what it makes. The clean re-run check reads the commands in the table below, so keep its format if you edit it. Version 2 adds the Scope 3 estimate, PCAF financed emissions and the exposure by fund; the method is in my-work/week-6/fund-level/library/methodology/scope3-pcaf-method-v1.md.
An analyst or IT colleague runs the commands below: they are programs, not spreadsheets. Sarah's part is the review at the end: open the dashboard, read the headline numbers against last year's, check the LP report, and decide which data requests to send. Nothing here needs Kith.
my-work/week-6/manifest-v*.json) lists every input and its fingerprint; a changed input needs a new manifest version.my-work/week-6/pipeline/questionnaire-reads-v1.csv: each row is an answer with a quote that the intake step checks word for word against the questionnaire.my-work/week-6/pipeline/bls-average-pay-v1.csv gives the URL of each) and check the spend map (my-work/week-6/pipeline/scope3-spend-map-v2.csv) still fits each company. Save changed files as new versions.my-work/week-6/fund-level/library/framework-freshness-v1.csv) and record what you read in the verification log.| Step | Command | What it makes |
|---|---|---|
| 1 | python3 -I my-work/week-6/pipeline/intake-v1.py . my-work/week-6/fund-level/case-northwood-cy2025-v2 | The case: data table (including Scope 3, financed emissions and the by-fund rows), settings, facts, company map |
| 2 | python3 -I my-work/week-6/fund-level/engine/link_register.py | Links each disclosure concept to the framework requirements |
| 3 | python3 -I my-work/week-6/fund-level/engine/run_engine.py --case my-work/week-6/fund-level/case-northwood-cy2025-v2 | The filled register, coverage matrix, framework content indices and the LP report text |
| 4 | python3 -I my-work/week-6/fund-level/engine/run_applicability.py --case my-work/week-6/fund-level/case-northwood-cy2025-v2 | Which frameworks apply: yes, no, unknown, monitor |
| 5 | python3 -I my-work/week-6/fund-level/qa/run_qa.py --case my-work/week-6/fund-level/case-northwood-cy2025-v2 | The 22 quality checks on the report; it must say QA PASS |
| 6 | python3 -I my-work/week-6/pipeline/build-d2-v1.py . | D2: evidence map, data requests, applicability and coverage files |
| 7 | python3 -I my-work/week-6/pipeline/build-d3-v1.py . | D3: the LP report folder (refuses to copy if QA failed or a company is identifiable) |
| 8 | python3 -I my-work/week-6/pipeline/build-d4-v1.py . | D4: the dashboard page, its data and the query pack |
| 9 | python3 -I my-work/week-6/pipeline/build-d5-v1.py . | D5: the ten board pages |
For a new year, use a new case folder and output version: use a new case folder whose name ends in the next version number and change the V setting at the top of the four build scripts. The engine and QA name their outputs after the case folder.
python3 -I my-work/week-6/fund-level/dashboard/check_dashboard.py . and python3 -I my-work/week-6/fund-level/board-pages/check_board_pages.py .; both must end in PASS.my-work/week-6/fund-level/tests/ to prove every check can still fail.my-work/week-6/fund-level/register/northwood-data-requests-v2.md and send the requests you decide to send (nothing is sent by the scripts).my-work/week-6/pipeline/narrative-v2.json. The scripts resolve every figure from the data; never type a figure into the text.Portfolio edition: Kith Climate cohort case, fictional client.