Validation Coverage

This document names the local validation commands that bind Harness rules to feedback loops for this Django target.

Normal Gate

Run the project wrapper before completing implementation or Harness-maintenance work:

.\.venv\Scripts\python.exe scripts\check_harness.py

On macOS or Linux, use the project virtual environment Python at .venv/bin/python when available.

scripts/check_harness.py runs:

• scripts/check_docs_drift.py
• scripts/check_structure.py
• scripts/check_encoding_hygiene.py
• scripts/check_effectiveness_plan.py --require-report
• scripts/check_failure_memory.py
• scripts/check_decision_memory.py
• manage.py check
• manage.py test

Focused Checks

Run these directly when working on their specific surface:

.\.venv\Scripts\python.exe scripts\check_docs_drift.py
.\.venv\Scripts\python.exe scripts\check_effectiveness_plan.py --require-report
.\.venv\Scripts\python.exe scripts\check_failure_memory.py
.\.venv\Scripts\python.exe scripts\check_decision_memory.py

Use Harness Doctor as a diagnostic, not as proof of agent effectiveness:

.\.venv\Scripts\python.exe harness-starter-kit\scripts\harness_doctor.py --target .

What This Does Not Prove

Passing local checks shows that the repository-visible Harness health gate is clean. It does not prove that agent effectiveness improved. Use docs/evaluation.md and effectiveness reports to record comparable task outcomes.