What problem does it solve?
Dutch government appointment news from the Algemene Bestuursdienst (ABD) is published as unstructured HTML, making it slow and error-prone to manually extract structured data for open government datasets like Polder. This skill automates that parsing process to turn raw ABD news articles into ready-to-use structured records.
Core Features & Use Cases
- Structured Proposal Generation: Parses ABD news HTML into Polder-compliant Membership-proposal objects with all required fields including person names, organizational hierarchies, role titles, dates and evidence snippets.
- Multi-Event Type Support: Handles appointments, resignations, contract extensions and obituaries, with tailored logic for each event type including automatic mandate closure for obituaries.
- Confidence Scoring: Assigns evidence-based confidence scores to each proposal based on explicit fact presence and external verification sources, with clear reasoning for review workflows.
- Use Case: A Polder dataset maintainer can process a new ABD news article in seconds to generate a validated, merge-ready proposal instead of spending minutes on manual data entry and validation.
Quick Start
Use the parse-abd-nieuws skill to process the ABD news HTML file at _cache/abd-nieuws/2026-05-09-marleen-heijster.html and output the structured membership proposals.