What problem does it solve?
Privacy decisions for clinical de-identification are often implicit and unrecorded, making pipelines hard to review and audit. This Skill turns five explicit policy choices into a reviewable, deterministic DEID-POLICY.md draft before any privacy work begins.
Core Features & Use Cases
- Bounded decision collection: Captures exactly five canonical choices (jurisdiction, recall floor, surrogate strategy, model policy, audit location) with strict normalization and rejection of free-form values.
- Deterministic template rendering: Fills a versioned local template so identical inputs produce byte-for-byte identical output, with atomic file replacement and symlink protection.
- Human approval gate: Writes a DRAFT status that requires explicit human review before the policy controls any de-identification run.
- Use Case: Before running a clinical text de-identification pipeline, a project owner answers the five policy questions and receives a DEID-POLICY.md draft that downstream skills apply, transform against, and audit.
Quick Start
Ask the assistant to set up an OpenMed de-identification policy for your project directory and answer the five decision prompts.
