CTC ↔ PSD Course Equivalency

Maps courses from 6 Washington community & technical colleges to PSD high school credit types per the WA State Board of Education 24-credit framework.

Live: https://psd401.github.io/ctc-psd-equivalency/

Audiences

• Counselors, CTC staff, students & families — read-only public view: course → HS credit value + type
• District deciders (Chief Academic Officer · Director of Secondary Teaching & Learning · Director of CTE · Director of Research & Assessment) — edit-enabled view; decisions persist to a Google Sheet via Apps Script

Two outputs from one data source

File	Audience	Edits
ctc-psd-equivalency.html + equivalency-data.json	Public	No
ctc-psd-decisions.html	Deciders (local-only, API-key gated)	Yes — saves to Sheet via Apps Script

Institutions covered

College	Catalog platform	Course count (2025-2026)
Tacoma Community College	Coursedog	828
Olympic College	Acalog	1,246
Pierce College	Acalog	947
Green River College	Acalog	1,378
Clover Park Technical College	SmartCatalog	1,116
Bates Technical College	Drupal	1,285
Total		6,800

Quick start

# Rebuild HTML + sidecar from existing classified dataset
python build_html.py

# Refresh one institution's catalog
python build_dataset.py olympic

# Refresh all enabled institutions
python build_dataset.py

# Re-classify the merged dataset in place (after editing classify_courses.py)
python classify_courses.py

# Re-merge per-institution files into the unified dataset
python merge_catalogs.py

# Year-over-year diff after annual catalog update
python diff_catalogs.py --year-from 2025-2026 --year-to 2026-2027 -o diff.md

Viewing locally

ctc-psd-equivalency.html (the public viewer) fetches its data from the equivalency-data.json sidecar, which the browser blocks under file://. Don't double-click it — serve it over HTTP instead:

./serve.sh        # serves this folder at http://localhost:8000
# then open http://localhost:8000/ctc-psd-equivalency.html

ctc-psd-decisions.html embeds its data inline and opens fine from file://.

See PIPELINE.md for the full architecture and operational guide.

OSPI-standards audit workflow

For each credit type, an LLM workflow compares course descriptions against the relevant WA OSPI K-12 Learning Standards and flags misclassifications. Roughly $0.05/course in API costs.

# Generate workflow script for one credit type
python audit_credit_type.py "Health"                    # all candidates
python audit_credit_type.py "CTE" --max-confidence 0.85 # focus on uncertain
python audit_credit_type.py "Math" --include-institutions tcc olympic

# Workflow runs externally (1 standards agent + N verdict agents in parallel)
# Then capture the result JSON and apply:
python apply_audit_decisions.py audit-health.json --dry-run
python apply_audit_decisions.py audit-health.json

Decisions posted by the audit are flagged decided_by = "Director of Research & Assessment (AI-assisted audit)" so they're distinguishable from human-made decisions. The Sheet is append-only, so any human override of an audit decision is preserved as a new row.

Completed audits: see audit-health-vs-ospi.md and audit-cte-vs-ospi.md.

Decisions backend setup

See decisions_setup/SETUP.md. The backing Google Sheet uses an append-only v2 schema:

decision_id | course_code | institution | applies_to | status |
override_credit_types | override_hs_credits | rationale | decided_by |
decided_date | source_citation | decided_for_year |
is_current | superseded_by | created_at | last_updated

WA Common Course Numbers (&-prefixed) default to applies_to=all, so a decision about HIST&146 at TCC automatically applies at every college offering it.

Deploy

./deploy.sh           # stages docs/ for GitHub Pages
git add docs/ && git commit -m "Deploy" && git push

License

Internal Peninsula School District tool. Not for redistribution.