Which Documentation Sections Should a Model Draft? A Free‑Tier Risk‑Split Workflow

Описание и применение

В workflow определяются, какие разделы документации могут быть сгенерированы моделью, а какие остаются под ответственностью человека. С помощью YAML‑файла и проверяющего скрипта гарантируется, что модель не попадает в зоны, требующие человеческого суждения.

Задача

Determining which documentation sections can be safely drafted by an AI model to reduce review effort while maintaining accountability and verifiability.

Результат

The workflow reduces review time by limiting AI drafting to fact‑extractable sections, enforces ownership boundaries with a YAML file and a verifier script, and operates entirely on free infrastructure.

Рабочий процесс

• Update ownership.yaml with human owners for non‑draftable sections
• Generate API reference and setup stubs from source to provide a skeleton
• Draft allowed sections with a free AI server, marking output with <!--model-drafted-->
• Run the verifier script to flag any model output in human‑owned sections and open a review listing blocked sections first

Доказательства

The table below is the core artifact of this workflow. Each row names a documentation section, marks whether a model may draft it, and states who owns the final text.

Источник

The companion script reads the ownership file and fails whenever a human‑owned section contains model output.

Источник